DEVELOPMENT OF AN OPTIMAL DRAGONFLY-LIKE

FLAPPING WING STRUCTURE FOR USE IN BIOMIMETIC

MICRO AIR VEHICLES

PRAVEENA NAIR SIVASANKARAN

THESIS SUBMITTED IN FULFILMENT OF THE

REQUIREMENTS FOR THE DEGREE OF DOCTOR

OF PHILOSOPHY

FACULTY OF ENGINEERING

UNIVERSITY OF MALAYA

KUALA LUMPUR

2015

Univers

ity of

Mala

ya

ii

UNIVERSITY OF MALAYA

ORIGINAL LITERARY WORK DECLARATION

Name of Candidate: (I.C/Passport No: )

Registration/Matric No:

Name of Degree:

Title of Project Paper/Research Report/Dissertation/Thesis (“this Work”):

Field of Study:

I do solemnly and sincerely declare that:

(1) I am the sole author/writer of this Work;

(2) This Work is original;

(3) Any use of any work in which copyright exists was done by way of fair

dealing and for permitted purposes and any excerpt or extract from, or

reference to or reproduction of any copyright work has been disclosed

expressly and sufficiently and the title of the Work and its authorship have

been acknowledged in this Work;

(4) I do not have any actual knowledge nor do I ought reasonably to know that

the making of this work constitutes an infringement of any copyright work;

(5) I hereby assign all and every rights in the copyright to this Work to the

University of Malaya (“UM”), who henceforth shall be owner of the

copyright in this Work and that any reproduction or use in any form or by any

means whatsoever is prohibited without the written consent of UM having

been first had and obtained;

(6) I am fully aware that if in the course of making this Work I have infringed

any copyright whether intentionally or otherwise, I may be subject to legal

action or any other action as may be determined by UM.

Candidate’s Signature Date:

Subscribed and solemnly declared before,

Witness’s Signature Date:

Name:

Designation:

Univers

ity of

Mala

ya

iii

ABSTRACT

Biomimetic Micro Air Vehicles (BMAV) are unmanned, micro-scaled aircrafts that are

bioinspired from flying organisms to achieve lift and thrust by flapping their wings.

Micro Air Vehicles (MAV) are a relatively new and rapidly growing area of aerospace

research. They were first defined by the US Defense Advanced Research Projects

Agency (DARPA) in 1997 as unmanned aircraft that are less than 15 cm in any

dimension. This allows BMAV to potentially be smaller and more lightweight than the

other two types. These characteristics make BMAV ideally suited for flight missions in

confined areas (e.g. around power lines, narrow streets, indoors, etc.). Therefore,

BMAV structural components must be ultra-lightweight, compact, and flexible. Most

past MAV research has focused on fixed wings, which are essentially scaled-down

versions of wings on conventional fixed wing aircraft. These wings are unsuitable for

BMAV due to their lack of flexibility. So a new type of structural wing design is

required for BMAV. In this work, a dragonfly wing structure is mimicked to construct a

new BMAV wing design. A dragonfly (Odonata) was selected for biomimicry, because

they are highly maneuverable flyers, capable of hovering, rapid forward flight, or

reverse flight. Therefore, structurally analyzing these wings could yield results that

inspire the design of more effective wings for BMAVs. The overall objective of this

research is to develop a simplified wing model for a BMAV, bioinspired from actual

dragonfly wings. A simplified model was created using spatial network analysis, a

topological optimization method. These simplified wing frame models were then

fabricated using seven different types of materials. Stainless steel type 321, balsa wood,

red pre-impregnated fiberglass, black graphite carbon fiber, polyvinyl acid, acrylic and

acrylo-nitirile butadiene styrene. These wing frame structures were fabricated using

laser cutting machine and a 3D printer. These wing frames were then immersed in a

chitin-chitosan membrane by a casting method. These wing frames were subjected to

Univers

ity of

Mala

ya

iv

mechanical testing’s such as bending and tensile to study its suitability for use in a

BMAV. A flapping mechanism was also created and used to produce flapping motion

on these BMAV wings and an actual dragonfly wing (for comparison). The aero elastic

properties of both the BMAV and actual dragonfly wings were examined using two

high speed frame camera. The bending angle, displaced distance or deflection, wing tip

angle, and the wing tip rotational twist speed were analyzed at the flapping frequencies

of 10,20, 30 Hz, 60 Hz and 120 Hz.

Univers

ity of

Mala

ya

v

ABSTRAK

Kenderaan udara mikro biomimetic merupakan sebuah jentera yang tidak memerlukan

pemandu dan bersaiz mikro. Inspirasi untuk mereka bentuk kenderaan seperti ini

diperoleh dari melihat serangga yang boleh terbang. Kenderaan udara mikro ini

merupakan teknologi yang baharu dan sedang berkembang di dalam bidang

aeroangkasa. Kenderaan mikro ini memporeleh definisi daripada US Defense Advanced

Research Projects Agency (DARPA) pada tahun 1997. Mereka mengkategorikan

kenderaan tanpa pemanudu ini haruslah mempunyai ukuran dimensi yang kurang

daripada 15 cm. Ini menjadikan BMAV kecil dan ringan berbanding kenderaan udara

yang lain. Ini juga membolehkan BMAV sesuai digunakan di kawasan yang kecil dan

sukar dilalui manusia ( kawasan yang dikelilingi kabel voltan tinggi serta lorong yang

sempit). Maka, komponene struktur BMAV mestilah ultra-ringan, padat dan fleksibel.

Kajian terdahulu memberi tumpuan kepada sayap yang statik yakni mengurangkan skala

kapal terbang yang sedia ada. Sayap seperti ini tidak sesuai bagi kenderaan udara mikro

disebabkan kurangnya fleksibiliti. Jadi, struktur sayap yang baharu direka untuk BMAV

ini. Di dalam penyelidikan ini, sayap pepatung telah dimimik untuk merekabentuk

sayap baru bagi BMAV ini. Spesis pepatung (Odonata) telah dipilih untuk dimimik

keranapepatung merupakan serangag terbang yang sangat efisien. Pepatung boleh

melakukan pelbagai aksi sewaktu terbang seperti terbang tanpa mengibas sayapnya

untuk waktu yang lama, terbang sambil menukar arah, atau terbang songsang. Objektif

penyelidikan ini adalah untuk merekabentuk sayap model sayap yang telah

disimplifikasikan untuk BMAV. Sayap pepatung menjadi model inspirasi kami. Sayap

pepatung ini disimplifikasikan menggunakan kaedah optimasi yang dinamakan “ spatial

network analysis’. Tujuh jenis bahan digunakan untuk menghasilkan model sayap ini.

Keluli tahan karat jenis 321, kayu balsa, fiberglass merah, carbon fiber, polyvinyl acid,

acrylic and acrylo-nitrile butadiene styrene. Sayap ini dihasilkan menggunakan mesin

Univers

ity of

Mala

ya

vi

laser dan printer 3D. Sayap ini kemudiannya disalut dengan membran chitin-chitosan.

Ujian tegangan dan lenturan dijalankan ke atas sayap ini. Sebuah mekanisme yang

memiliki kebolehan mengimbas disediakan bagi menguji kebolehan sayap-sayap yang

dihasilkan ini. Bagi membandingkan kebolehan sayap-sayap ini, sayap pepatung yang

sebenar juga diuji di atas mekanisme ini. Kebolehan aero-elastik sayap sayap ini diukur

dan dibandingkan dengan sayap pepatung yang asal menggunakan dua kamera halaju

tinggi. Ukuran seperti sudut lenturan, jarak imbasan, sudut hujung sayap, sudut putaran

hujung sayap dan halaju putaran hujung sayap diuji pada frekuensi 10, 20, 30, 60 dan

120 Hz.

Univers

ity of

Mala

ya

vii

DEDICATIONS

this thesis is dedicated especially

to my beloved father, mother, and husband

Mr. Sivasankaran Nair, Mrs. Santhakumari and Mr. Rajendra Nath

The caring ones,

Kishan Nair Sivasankaran, Rubentheren Viyapuri

Respected supervisors,

Dr. Thomas Arthur Ward

Associate Professor Dr. Mohd Rafie Johan

Thanks for all the support.

My love for you all remains forever.

Univers

ity of

Mala

ya

viii

ACKNOWLEDGEMENT

I take this opportunity to express my profound gratitude to my supervisor Dr.

Thomas Arthur Ward for his exemplary guidance, monitoring, patience and constant

encouragement throughout the duration of my studies. I would also like to

acknowledge with much appreciation the supportive role of my co-supervisor Associate

Professor Dr. Mohd Rafie Johan for guiding me professionally throughout the

candidature.

I would like to express the deepest appreciation to all my lab members Mr.

Rubentheren Viyapuri, Mr. Christopher Fearday and Mr. Erfan Salami. I’m thankful for

being blessed with a friendly and cheerful group. They were always there to lend me a

helping hand under many situations. I wish to take this opportunity to express my

heartfelt gratitude to all of them. I also take this opportunity to express a deep sense of

appreciation to lab assistants Mr. Fauzi who have helped me during my lab experiments.

Special thanks to University Malaya for sourcing me a quality workstation.

Of course no acknowledgments would be complete without giving thanks to my

family. I would like to thank my family for their unconditional support. In particular,

the motivation and encouragement given by my husband has helped me pull through

many critical situations throughout my candidature. Also, the absolute love and

blessings showered by my mum, dad, and brother. A special thanks to Rubentheren

Viyapuri for being a good friend in need.

Above all, I thank God for his grace, wisdom, favor and protection. I could

never have done this without the faith I have in you, Baba.

There are so many others whom I may have inadvertently left out and I sincerely

thank all of them for their help.

Univers

ity of

Mala

ya

ix

TABLE OF CONTENTS

PAGE

ABSTRACT iii

DEDICATION vii

ACKNOWLEDGEMENT viii

TABLE OF CONTENT ix

LIST OF FIGURES xiv

LIST OF TABLES xvii

LIST OF SYMBOLS AND ABBREVIATIONS xviii

LIST OF APPENDICES xxii

CHAPTER 1: INTRODUCTION 1

1.1 Biomimetics 1

1.2 Unmanned aerial vehicles (UAVs) 1

1.3 Problem statement 3

1.4 Objectives 5

1.5 Procedures 7

1.6 Outline of thesis 7

CHAPTER 2: LITERATURE REVIEW 9

2.1 Introduction 9

2.2 Insect flight 9

2.3 Dragonflies 10

2.4 Spatial network analysis 17

Univers

ity of

Mala

ya

x

2.5 Fabrication of artificial insect wings 19

2.6 Frequencies of insect wings 21

2.7 Static test conducted on real and artificial wings 22

2.8 Flapping mechanism 23

2.9 Ornithopters 23

2.10 High speed camera imaging technique 24

CHAPTER 3: RESEARCH METHODOLOGY 26

3.1 Introduction 26

3.2 Wing model overview 26

3.3 Spatial network analysis 28

3.3.1 Canny edge detection algorithm 28

3.3.2 Proximity index 31

3.3.3 CAD model 34

3.4 Finite element analysis 35

3.4.1 Element and mesh 37

3.4.2 Modal analysis: Mode shapes and MAC 39

3.5 Artificial wing frames and nano-composite chitosan membrane 41

3.5.1 Fabrication of stainless steel (Type 321) wing frame s 40

3.5.2 Fabrication of balsa wood wing frames 42

3.5.3 Fabrication of black graphite carbon fiber wing frames 42

3.5.4 Fabrication of red prepreg fiberglass wing frames 44

3.5.5 Fabrication of polylactic acid (PLA) wing frames 45

Univers

ity of

Mala

ya

xi

3.5.6 Fabrication of acrylic wing frames 45

3.5.7 Fabrication of acrylonitrile butadiene styrene ABS wing frames 46

3.5.8 Chitosan nano-composite solution (membrane) 47

3.6 Numerical bend-twist coupling and static strength simulation

analysis 48

3.6.1 Numerical bend-twist coupling analysis 48

3.6.2 Static strength simulation analysis 48

3.7 Experimental set up 49

3.8 Flapping mechanism 53

CHAPTER 4: RESULTS AND DISCUSSION 55

4.1 Introduction 55

4.2 Simplified model creation 55

4.3 Mode shape analysis and MAC 59

4.4 Numerical bend-twist coupling analysis 64

4.5 Tensile test simulation result 66

4.5.1 Tensile simulations (wing frames without membranes) 66

4.5.2 Tensile test simulations results (wing frames with membrane) 71

4.5.3 Bending test simulation results (wing frames without membrane) 73

4.5.4 Bending test simulation results (wing frames with membrane) 75

4.5.5 Tensile test experimental results 76

4.5.6 Bending test experimental results 84

4.6 High speed camera image results analysis 87

Univers

ity of

Mala

ya

xii

4.6.1 Bending angle versus flapping frequency 92

4.6.2 Wing tip deflection versus flapping frequency 95

4.6.3 Wing twist angle versus flapping frequency 97

4.6.4 Wing tip twist speed versus flapping frequency 99

CHAPTER 5 – CONCLUSION 102

5.1 Summary 102

5.2 Future work 106

REFERENCES 107

LIST OF PUBLICATIONS 119

APPENDICES 121

Univers

ity of

Mala

ya

xiii

LIST OF FIGURES

PAGE

Figure 1.1 Illustration of the overall procedures in developing a

BMAV wing frame 7

Figure 2.1 Dragonfly species used in this study 11

Figure 2.2 Strokes of wings during flapping motion 11

Figure 2.3 The front view of a dragonfly in flight 15

Figure 2.4 Schematic diagram of Gui et al (2010) of fire ant alate

wing motion 25

Figure 3.1 Main parts of a dragonfly wing 27

Figure 3.2 Spatial Network Model 33

Figure 3.3 (a) Digital image of the forewing of a dragonfly;

b) forewing created by Canny edge algorithm;

c) Canny edge forewing created after noise

minimization and main pattern identification;

d) forewing CAD model 34-35

Figure 3.4 Stainless steel type 321 wing frames; a) forewing,

b) hindwing 41

Figure 3.5 Balsa wood wing frames; a) forewing, b) hindwing 42

Figure 3.6 Waviness of carbon fiber yarns 43

Figure 3.7 Black graphite carbon fiber wing frames; a) forewing,

b) hindwing 43

Figure 3.8 Red pre-preg fiberglass wing models after immersion

in chitosan nanocomposite solutions;

a) forewing, b) hindwing 44

Figure 3.9 PLA wing models after immersion in chitosan

nano-composite solutions; a) forewing, b) hindwing 45

Figure 3.10 Acrylic wing models after immersion in chitosan

nano-composite solutions; a) forewing, b) hindwing 45

Figure 3.11 ABS wing models after immersion in chitosan

nano-composite solutions; a) forewing, b) hindwing) 46

Figure 3.12 Wing structure immersed in chitosan nanocomposite

solution; (a) fore wing, (b) hind wing 47

Univers

ity of

Mala

ya

xiv

Figure 3.13 Placement of constraints fixed in a natural frequency

analysis 48

Figure 3.14 Constraints for a static strength analysis 49

Figure 3.15 INSTRON Universal testing machine 51

Figure 3.16 (a) Actual experimental set up of the high speed camera

imaging technique, (b) Schematic diagram of the

experimental set up 51-52

Figure 3.17 Front view (a) and side view (b) of the wing motion

captured and measurement axes. 52

Figure 3.18 Flapping mechanism used in this research 54

Figure 4.1 Fabricated simplified model (a) forewing;

(b) hindwing 55

Figure 4.2 Typical dragonfly wing membrane pattern arrangement 57

Figure 4.3 Example of connecting strong edges using cloud points 58

Figure 4.4 Simplified models, (a) forewing, (b) hindwing 59

Figure 4.5 Similar mode shapes and corresponding natural

frequencies for both detailed (left) and simplified

models (right) : forewing 60

Figure 4.6 Similar mode shapes and corresponding natural

frequencies for both detailed (left) and simplified

models (right) : hindwing 61-62

Figure 4.7 The static bend-twist coupling graph of both detailed

and simplified models; forewing 64

Figure 4.8 The static bend-twist coupling graph of both detailed

and simplified models; hindwing 65

Figure 4.9 Von Misses stress simulation results of different forewing

frame structures; (a) stainless steel (Type 321),

(b) balsa wood, (c) black graphite carbon fiber,

(d) red prepreg fiberglass e) PLA, f) acrylic, g) ABS 68

Figure 4.10 Von Misses stress simulation results of different forewing

frame structures; (a) stainless steel (Type 321), (b) balsa

wood, (c) black graphite carbon fiber, (d) red prepreg

fiberglass, e) PLA, f) acrylic, g) ABS 69

Univers

ity of

Mala

ya

xv

Figure 4.11 von Misses stress simulation results of different

forewing model structures; (a) stainless steel (Type 321),

(b) balsa wood, (c) black graphite carbon fiber,

(d) red prepreg fiberglass, (e) PLA, (f) acrylic, (g) ABS 71

Figure 4.12 von Misses stress simulation results of different

forewing model structures; (a) stainless steel (Type 321),

(b) balsa wood, (c) black graphite carbon fiber,

(d) red prepreg fiberglass, (e) PLA, (f) acrylic, (g) ABS 72-73

Figure 4.13 von Misses stress simulation results of different forewing

model structures; (a) stainless steel (Type 321),

and (b) balsa wood 74

Figure 4.14 Von Misses stress simulation results of different hindwing

model structures; (a) stainless steel (Type 321),

and (b) balsa wood 74

Figure 4.15 von Misses stress simulation results of different forewing

model structures; (a) stainless steel (Type 321),

and (b) balsa wood 75

Figure 4.16 von Misses stress simulation results of different hindwing

model structures; (a) stainless steel (Type 321),

and (b) balsa wood 75

Figure 4.17 Tensile stress vs tensile strain (engineering) of all four

forewing frame structures ; a) stainless steel (Type 321),

b) balsa wood, c) black graphite carbon fiber,

d) red prepreg fiberglass, e) PLA, f) acrylic and g) ABS 80-82

Figure 4.18 Tensile stress vs tensile strain (engineering) of all four

hindwing frame structures ; a) stainless steel (Type 321),

b) balsa wood, c) black graphite carbon fiber,

d) red prepreg fiberglass, e) PLA, f) acrylic and g) ABS 82-83

Figure 4.19 Comparison stress vs comparison strain graphs of forewing

wing models; a) stainless steel (Type 321) ,b) balsa wood 85

Figure 4.20 Comparison stress vs comparison strain graphs of hindwing

wing models; a) stainless steel (Type 321) ,b) balsa wood 86

Figure 4.21 Degrees of freedom for the wings of flying insects 88

Figure 4.22 Side view of the dragonfly flapping wing (gray scale)

captured by the high speed camera during one flapping

cycle at 30Hz. a) start of downstroke, b) mid-downstroke,

c) end of downstroke, d) start of upstroke, e) mid-upstroke,

f) end of upstroke 89

Univers

ity of

Mala

ya

xvi

Figure 4.23 Front view of the dragonfly flapping wing (gray scale)

captured by the high speed camera during one flapping

cycle at 30Hz. a) start of downstroke, b) mid-downstroke,

c) end of downstroke, d) start of upstroke, e) mid-upstroke,

f) end of upstroke 90

Figure 4.24 Bending angle of different wing frames;

(a) without membrane and (b) with membrane 94

Figure 4.25 Wing tip deflection of different wing frames;

(a) without membrane and (b) with membrane 96

Figure 4.26 Wing twist angle of different frames versus flapping

frequency; (a) without membrane and (b) with membrane 98

Figure 4.27 Wing tip twist speed of different frames versus flapping

frequency; (a) without membrane and (b) with membrane 100

Univers

ity of

Mala

ya

xvii

LIST OF TABLES

PAGE

Table 3.1 Properties of Teflon 38

Table 3.2 Mechanical properties of frame structure materials 40

Table 3.3 Specifications of black graphite carbon fiber 43

Table 4.1 Model Specifications in Terms of Number of Membranes,

Density and Patterns 56

Table 4.2 MAC values for all five compared mode shapes for both

forewing and hindwing 63

Univers

ity of

Mala

ya

xviii

LIST OF SYMBOLS AND ABBREVIATIONS

Symbols and Abbreviations Full name

UAV Unmanned aerial vehicle

MAV Micro air vehicle

BMAV Biomimetic micro air vehicle

Hz Hertz

PLA Polylactic acid

ABS Acrylonitrile-butadiene styrene

dpi Dots per inch

m Meter

μ Micro

FEA Finite element analysis

ω circular frequency

θ Bending angle

d Displaced distance or deflection

α Wing tip angle

CAD Computer aided design

3D three dimensional

mm millimeter

Univers

ity of

Mala

ya

xix

MPa Mega Pascal

GPa Giga Pascal

Univers

ity of

Mala

ya

xx

LIST OF APPENDICES

PAGE

Appendix A Force displacement graph of dragonfly fore wing 121

Appendix B Force displacement graph of dragonfly hind wing 123

Appendix C Force displacement graph of BMAV fore wing 126

Appendix D Force displacement graph of BMAV hind wing 128

Univers

ity of

Mala

ya

1

CHAPTER 1: INTRODUCTION

1.1 Biomimetics

Biomimetics or biomimicry is a concept of imitating elements of nature to solve or

to ease complex human tasks. The terms biomimetics and biomimicry come from Ancient

Greek: biosmimesis or biosmios which means imitating life (Julian et al. 2006). The

development of biomimetics has created new bio-technological solutions at macro and

nanoscales. The application of biomimetics can be seen in many fields from cars to

computers. Optimization methods such as swarm intelligence and artificial neural networks

were inspired by biomimetics. There are three areas in where biotechnological solutions

can be applied (Rhett Butler, 2015): Replicating natural production methods (chemical

compounds from plants), Mimicking mechanisms found in nature (dragonfly, Velcro) and

finally observation of social behavior of organisms (ants and bees).

1.2 Unmanned aerial vehicles (UAVs)

The Unmanned Aerial Vehicle or commonly known as UAV, is an aircraft with no

pilot on board. UAVs can be controlled by a ground control station, routed by pre-

programmed flight plan or fitted with a complex automation system. UAVs are currently

used for a number of missions, including reconnaissance and attack roles.

Micro air vehicles (MAV) are a new type of unmanned air vehicle which is

significantly smaller than a UAV. Defense Advanced Research Projects Agency defines a

MAV as an aircraft which has a size less than 15 cm in length, width or height. This puts

MAV smaller than any UAV developed to date. MAVs should be thought of as aerial

robots, whose mobility will allow them to be deployed to remote or hazardous locations to

perform a variety of missions which includes surveillance, targeting and even rescue

Univers

ity of

Mala

ya

2

missions. (Michelson et al, 2002). There are different types of MAVs such as fixed wing

models, (FMAVs) rotary wing models (RMAVs) and insect-like (flapping wing) models,

(BMAVs). FMAVs are micro-scaled air vehicles with fixed (stationary) wings that

generates lift but not thrust. There is a separate system attached (propulsion) to generate

thrust. They are used in non-confined spaces because they should maintain a continuous

forward velocity in order to generate lift (wings). Rotors are used to achieve lift and thrust

for RMAVs. Since they are able to hover they are much efficient than FMAV. However,

the size is their limitation. RMAVs are generally larger than FMAVs or BMAVs. BMAVs

are micro or nano-scaled aerial vehicles biomimicked from a biological organism. BMAVs

are the smallest MAVs designed and they have flexible flapping wings. Due to their small-

size range, the BMAVs are able to travel in confined spaces. According to James and

Michael (1997) the system components in a MAV should be integrated and possess

multifunctional characteristics to be able to overcome the challenge faced in designing an

aircraft with constrained dimensions. A MAV may only weigh about 50 grams yet; MAVs

must be capable of hovering as long as 20 to 60 minutes while carrying a load of 20 grams

a distance of approximately 10 km. It is essential to find the suitable materials and the

optimum structural balance for a MAV (James and Michael, 1997).

1.3 Problem statement

The study of biomimetics as mentioned earlier has motivated research into a new

class of micro-sized unmanned aircraft called Biomimetic Micro Air Vehicles (BMAVs).

Due to their small size and weight, BMAVs would be ideal for flying indoors or in

enclosed spaces. The primary payloads envisioned for a BMAV are ultra-lightweight,

compact electronic and surveillance detection equipment. Their miniature size makes them

difficult to detect, easy to quickly deploy by a single operator and relatively inexpensive to

fabricate, and allows the potential to fly them inside buildings or compact spaces. BMAVs

Univers

ity of

Mala

ya

3

are envisioned for use on civil and military missions that are of a limited duration, such as:

remote sensing of hazard sites (e.g. chemical spill, radiation, high voltage power lines etc.),

indoor video mapping, and police or military surveillance.

Some research has been conducted to biomimic the wing structure of dragonflies.

TU Delft researchers have developed the DelFly Explorer, a low weight (20 grams) MAV

which flaps and avoid obstacles autonomously (Michelson et al, 2002). Festo created

BionicOpter, a fully functional robotic dragonfly which is able to perform the aerial

acrobats of an actual dragonfly. BionicOpter is much larger than an actual dragonfly with a

weight of 175grams (TA Ward et al, 2015). Dragonflies (Odonata) have a highly

corrugated wing structure that consists of complex, wing membrane patterns (e.g.

quadrilaterals, pentagons and hexagons). Each pattern contributes to the wing’s flexibility

and stiffness, which determine its ability to withstand deformations. Although some

research has been conducted to biomimic the wing structure of dragonflies. (Tu Delft, 2013

and Yirka, 2013), none of these works biomimicked the actual wing structures of

dragonfly. Although other insect wings (e.g. cicada, beetle, etc.) have been replicated,

dragonfly wing structures are very complex making this difficult. Dragonfly wings are

highly corrugated in pattern, with differing thickness and tubular structures. Replicating an

exact copy is not a practical approach, due to limitations in fabrication methods. So a

simplified model is needed that takes these limitations into consideration. Like the wings of

a flying insect, the artificial wings of a BMAV not only must be flexible but also strong

enough to endure the aerodynamic forces produced by flapping motion. During flight, the

wings undergo significant bending and twisting deformations that can alter the direction

and magnitude of the aerodynamic forces generated (Combes S, 2003, 2010, Shang J.K et

al, 2009)

Univers

ity of

Mala

ya

4

Several fabrication methods for small insect-scale artificial wings have been

proposed. Combes and Daniel (2010) measured the wing flexibility of several insects and

found that the spanwise flexural stiffness was one to two orders of magnitude larger than

the chordwise flexural stiffness. The scope of their investigation was limited due to the

diversity in venation as well as the complex cross-sectional and planform geometries of the

insect wings. In contrast, the morphology and materials of artificial wings can be

manipulated to understand the effect of these properties on wing flexibility. Tanaka and

Wood (2010) investigated the effects of flexural and torsional wing flexibilities on lift

generation in hoverfly flight using an insect-scaled mechanical model of an artificial wing.

There are several past research studies that focus on insects such as hoverfly, beetle and

humming (Tanaka et al. 2010, Nguyen et al. 2010 and David et al. 2014). Also, most past

literature involves wing membrane materials (Bao, 2011, Ko, 2012, Joong, 2013). Very

little has been written about the BMAV wing frame structures that encase the membrane.

Several past research publications have been conducted on flying insect wing structures to

understand their aeroelastic properties. Ward et al. (2015) conducted a review on the

percentage of flying organisms studied; hawkmoth (44%), dragonfly (23%), beetle (10%),

butterfly and humming bird (7%), bat (4%), fruit fly (3%), honey bee (2%) and damselfly

(1%). Herbert et al. (2000) conducted numerical investigations on a tethered desert locust

(Schistocerca gregaria). They concluded that the wings must undergo an appropriate

aeroelastic wing deformation (through the course of a wing beat) in order to achieve an

efficient aerodynamic flow suitable for lift and thrust generation. Several studies showed

that flexible wings, capable of changing their camber, generate higher peak lift forces than

rigid wings (Mountcastle and Young, 2009).

Wing flexibility also prevents small tears or warping from occurring. Newman et al

(1986) suggested that dragonfly wings appear to be adapted for reversible failure in

Univers

ity of

Mala

ya

5

response to excess loads, enabling them to avoid permanent structural damage. Most past

research examined the effects of wing flexibility on aerodynamic performance by either

using numerical models or experimentation. However, very few researchers have attempted

to mimic the detailed structure of an actual insect wing. Hence, the overall objective of this

research would be to create a simplified dragonfly wing model for use in a BMAV. This

wing model is envisioned to be able to match the performance of an actual dragonfly wing.

1.4 Objectives

The overall objective of this research is to model a simplified dragonfly wing that

matches the performance and capabilities of an actual dragonfly wing. Firstly, a simplified

model of the complex corrugated dragonfly wing is created. As previously stated, this

model has a simplified design to suit appropriate machining tolerances. A mode analysis

and static bend-twist coupling studies were made on computer simulations of these models.

Secondly, these simplified models were fabricated using different materials. Simulations

and tensile and bending experimental tests were done to determine the most suitable

material. Thirdly, the most promising materials were attached onto a flapping mechanism

and their aero-elastic properties were measured. These were compared with an actual

dragonfly wing for comparison purposes. Listed below is a summary of the objectives

performed: analysis of dragonfly-inspired BMAV wing structures (e.g. patterns, veins, and

tubules of dragonfly wings) includes designing a simplified wing structure optimized using

spatial network analysis, conducting finite element analysis including modal analysis and

other mechanical simulation testing on the simplified design (static bend-twist coupling),

fabricating simplified wings using different materials (e.g. steel, balsa wood, carbon fiber,

red fiberglass, acrylic, ABS, and PLA), conducting mechanical testing on fabricated wings

to ensure it matches the capability of an actual dragonfly wing and finally to test the

Univers

ity of

Mala

ya

6

fabricated wings (chosen material) on a flapping mechanism to assess the performance of

each design.

1.5 Procedures

Figure 1.1 below shows the summary of the procedures involved in the fabrication

of these wings.

Figure 1.1: Illustration of the overall procedures in developing a BMAV wing frame

Figure 1.1 describes the overall steps involved in this research. It starts with a

comprehensive literature review which includes a research of dragonflies and BMAV in

general and research of dragonfly wing structures specifically. The second step involves

design and modeling of simplified wing structures using spatial network analysis, FEA

(finite element analysis), modal analysis, and static bend twist analysis. The third step

involves fabrication of the wings using different method (manual carving, laser machining

and 3D printer) and finally testing these fabricated wings on a flapping mechanism.

Univers

ity of

Mala

ya

7

1.6 Outline of thesis

There are five chapters in this thesis, and a list summarizing them is as stated. The

first chapter introduces BMAV and their potential applications. It provides an overview of

this research and lists the primary objectives. It also lists layout of the procedure for

accomplishing these objectives, the second chapter reviews past researches done with a

similar objective. It gives a brief review of past research done on dragonflies and other

related topics that subsequently contributes to this research. Other review on related topics

include mechanism of insect flight, simulation of insect flights, flapping mechanism

created by other researchers, finite element analysis studies done on insect flight,

fabrication of artificial wing frame and membranes of insect and the high speed camera

imaging technique. In the third chapter, experiments are discussed extensively. Sufficient

details are given so that future readers can replicate this work. It discusses the topological

optimization method used which is spatial network analysis, the finite element analysis

options chosen and the mode analysis study made for both detailed and simplified wings,

the mode shape comparison, the static bend-twist study conducted, the materials used for

the fabrication of the simplified wing models, the simulation and experimental tests done

on the fabricated wing frames and, finally, the high speed camera technique used to capture

the parameters necessary for an aero-elastic study, The fourth chapter discuss and assesses

all the data acquired from the experiments. Where possible, these results are compared with

data collected by other researchers, and the last chapter summarizes and concludes this

research. It lists the experience and challenges faced while performing these studies and

also suggest improvements that can be done in future studies.

Univers

ity of

Mala

ya

8

CHAPTER 2: LITERATURE REVIEW`

2.1 Introduction

This chapter highlights several subtopics that were studied to support this research.

The chapter introduces insect flight concepts, the wing structure of dragonflies, the spatial

network analysis, the type of artificial wings fabricated in previous insect studies, testing

conducted in previous studies, types of flapping mechanisms, and data collected from

previous high frame rate imaging system studies.

2.2 Insect flight

The size of insect wings is not an indication of the insects’ ability to fly. Flies and

certain types of ants (Dobsonflies and Antlions) with large wings are poor fliers when

compared to wasps and bees that have smaller wings (Smithsonian, 2010). Flight in insects

enabled them to protect themselves from danger, as a means of survival (food hunting),

reproduction, and to search for new habitats (Monarch Butterflies) (Rose, n.d)

Wing flapping patterns of insects can vary significantly from one species to another.

The African Grasshopper, has the ability to fly hundreds of miles in search of food. This is

mainly attributed to their complex wing structure (“Smithsonian”, 2010). Wings typically

make up approximately 1%-2% of an insect’s total body mass. In dragonflies, they must

flap a million times in an average lifespan of an insect but not without enduring collisions,

torsional loads and other various forms of deformation (Kutsch, 2002).

2.3 Dragonflies

Separate muscles control the fore and hind wings, making the phase relation

between the 2 parts as a distinctive feature of the dragonfly’s wing movement. For an

example, both parts tend to beat closer in phase during takeoff but out of phase when

Univers

ity of

Mala

ya

9

hovering. A reduction in oscillation is the main reason the dragonfly varies the phase for

different maneuvers. The relatively close distance between the fore and hind wings, about a

wings width apart, allows both parts to interact aerodynamically (Jane, 2004).

Dragonflies have shown different flight styles. The four different flight styles

observed in dragonflies are (Norberg, 1975): Counter stroking (fore and hind wings move

up and down about 180 degrees non-symmetrically), phase stroking (hind wings cycle

about 90 degrees before the fore-wings), synchronized stroking (fore and hind wings move

in unison) and gliding

Figure 2.1: Dragonfly species used in this study

Of the 4 different styles, the counter stroking is the most used when they are either

hovering or flying very slowly. This is both a powerful and efficient way of flying and

generates a lot of lift (Rüppell, 1989).

When flying at average speeds, phase stoking is the preferred style. This method

generates more thrust but provides less lift when compared to the counter stroke. In order to

maximize thrust or to change directions fast, the synchronized stroke is employed by the

Univers

ity of

Mala

ya

10

dragonfly. Finally, gliding where three different types of gliding have been recognized

(Reavis et al, 1988); free gliding ; where the insect stops stroking and glides slowly down,

updraft gliding at hill crests ; where the insect adjusts its wing position to float in the air

without the need to beat its wings, gliding in towed females ; where the female insect in the

wheel position holds her wings out and glides while the male provides the motion force.

(This is during the mating position). The muscles used to flap the wings are attached to the

wing base. Elevated muscle temperature increases flapping efficiency. For this reason the

dragonfly spends a lot of time and energy to maintain their flight muscles at elevated

temperatures (Reavis et al. 1988, Saharon et al. 1987). The dragonfly thorax appears

skewed when at rest, but at flight, the head is held low and the stroke of the wings are

parallel to the long axis of the flight muscles. This is done to improve mechanical

efficiency. During each stroke, smaller controller muscles (operating at the base of the

wings) adjust the wing shape and angle of attack. Four different thrust generating

mechanics may be employed by the dragonfly (Sun, 2002). The four different mechanisms

used by the dragonfly are: classical lift, supercritical lift, vortices and vortex shedding.

When the attack angle of the wing passes a critical value, supercritical lift occurs.

Similar to hovering, a high lift is generated over a short distance after which the wing stalls.

Dragonflies can maintain this position continuously by using short wing strokes. The use of

vortices and shed vortices in insect flight by Usherwood et al. (2002) was explored. Both

the movement of the wings through the air and the twisting of the wings at the end of each

stroke to generate thrust.

Dragonfly wings are not simple planar objects but complex dynamic structures. The

corrugations in the wings holding an aero-foil of air around the wings have the following

effects: Lower friction, wings are able to flex around several axes, and being able to

respond to both muscle actions and to inertia effects. The pterostigma on the leading edge

Univers

ity of

Mala

ya

11

near the tip of the wings are weights that improve the aerodynamic efficiency by causing

the wing tip to flex during a wing stroke. Dragonflies can fly with asymmetrical wing

stokes. This is achieved by varying what their wings are doing in a coordinated fashion.

They have to master to adjust wing shape, stroke length, angle of attack, stopping either

wings independent of the other and adjusting the relationship between any two wings on

either side of the body are among a few of the many subtle adjustments made in order to

maintain their flight style (Ellington, 1984 (a),(b) and 1985).

Dragonflies have a natural frequency range of 120-200 Hz and a wingbeat

frequency of around 30Hz. The flapping of the wings creates a whirlwind surrounding the

wing area. A dragonfly uses a rowing motion along an inclined stroke plane. When

hovering, the body lies almost horizontal while the wings push backward and downward

with the addition of the feathers slicing upwards and downwards at the end of each stroke

(Grodnitsky, 1999). In contrast to the dragonflies asymmetrical rowing motion, many other

insects use symmetrical back and forth strokes near a horizontal stroke plane. This type of

motion also allows a dragonfly to support much of its weight by the upward drag created

during the down-stroke. The flapping motion of an actual dragonfly is shown in Figure 2.2.

Univers

ity of

Mala

ya

12

Figure 2.2: The front view of a dragonfly in flight.

From observation and inspection on high speed films, dragonfly wings are found to

have plastic like features. Observations from high speed films show that the torsional wave

propagates from the wing tip to the root during pitch reversal. This would be in the

opposite direction if the muscles were actively pitching the wing, as one would expect the

propagation to start from the root. Aerodynamic forces and wing inertia are responsible for

pitching the wing in this tip-to-root direction (Ellington, 1999). The aerodynamic torque

and inertial force associated with the observed wing motions can be computed and

confirmed that they are sufficient to pitch the wings of dragonflies and other observed

hovering wing motions. In order to simplify control and save energy, insects can take

Univers

ity of

Mala

ya

13

advantage of the natural swinging motion near the end of its wing stroke (Lehman et al,

2005).

Dragonflies have a very respectable power to weight ratio. (The dragonfly has a

mass of less than 0.028 kg and can reach a speed of 1.6 km/h). This allows them to

accelerate up to 4g in a straight line and up to 9g in turns. This has been documented in

high speed videotapes of free-flying dragonflies as they pursue prospective prey or when

trying to break off attacks. The agility of a dragonfly is the reason it gets much interest

from BMAV researchers (Luttgess, 1989).

A dragonfly achieves efficient structural performance because of the variations on

its wing pattern. Its ability to withstand deformation can be attributed to the wings

flexibility and stiffness. The highly corrugated wing pattern structure is made up of

quadrilateral, pentagonal, and hexagonal shapes. Studies have shown that these wing

patterns promote rigidity and flexibility along the span of the wing allowing for efficient

use of energy required for flight. Conclusions generated by Maybury et al. (2004) are: The

wing patterns in general follow the tensile forces exhibited on the wing and the variation in

shape determines the amount of flexibility and stiffness in that specific area of the wing

For an example, the more rigid and stiff portions of the wing are the quadrilateral

shaped areas, while the more compartmentalized hexagonal areas are regions where

bending and swaying take place (Figure 3.1). Also, the degrees at which cells are free to

bend are determined by the connections between the cells. Two main types of joints occur

in the dragonfly wings, mobile and immobile. Some longitudinal veins are elastically

joined with cross veins, whereas other longitudinal veins are firmly joined with cross veins.

Scanning electron microscopy reveals a range of flexible cross-vein and main-vein

junctions in the wing, which allows local deformations to occur (Miller et al. 2004). The

Univers

ity of

Mala

ya

14

occurrence of resilin, a rubber-like protein, in mobile joints enables the automatic twisting

mechanism of the leading edge (Akira et al. 1988).

2.4 Biomimetic Micro Aerial Vehicles (BMAVs)

BMAV are micro or nano-scaled aerial vehicles. These vehicles are biologically-

inspired from flying insects, birds and bats to achieve lift and thrust by flapping their

wings. BMAV are lightweight and highly maneuverable. They are capable of flying in

confined areas, rubbles and even indoors. The flapping wings are able to produce a much

larger lift compared to a fixed wing. There are many technical challenges involved with

designing BMAV such as modelling and unsteady aerodynamics, compact flapping

mechanisms, ultra-lightweight materials and structures and importantly an ultra-lightweight

power supply. Because of all these factors, there are currently no operational BMAVs but

only research based prototypes. Researchers from Delft University of Technology have

published numerous articles on their BMAV called Delfly (Tu Delft, 2009). This BMAV

only weighs 3 g, and a wing span of 10 cm.

Figure 2.3. Delfly from TuDelft

There are numerous articles published on BMAVs. Various categories are studied

under the BMAV field which includes aerodynamics (47%), propulsion and mechanisms

Univers

ity of

Mala

ya

15

(22%), system design and guidance and control contributes 11% and the studies of

structures and materials which is 9% of the total. An ultra-light weight, micro-sized

mechanism and power supply to enable free flight is the most challenging task in

developing a BMAV. Madangopal et al. (2005) analyzed the energetics of a BMAV design

with a four-bar flapping mechanism. Conn et al. (2007) created a parallel crank-rocker

mechanism that allows unconstrained, integrated flapping and pitching motion. Bolsman et

al. (2009) created an actuation mechanism that reduces energy expenditure and amplifies

flapping amplitude. Guo et al. (2012) created an actuated flapping wing rotor model using

an piezoelectric actuator and did a comparison with their numerical model simulation.

The structures and materials of BMAV are often inspired from flying organisms,

such as insects or birds or even bats. Song et al. (2004) made a comprehensive analysis on

the mechanical properties of the forewing of the cicada. This serves as a baseline for

BMAV materials. Dirks and Taylor (2012) showed that the cross veins of a locust wing

provides a barrier against crack propagations. A simplified insect wing model (inspired by

a dragonfly) was modeled by Sivasankaran and Ward (2015) based on spatial networking

analysis. Rubentheren et al. (2015) produced a chitosan membrane to biomimic a dragonfly

wing membrane and applied them to simplified wing frame structures.

2.5 Fabrication of artificial insect wings

Hisayoshi et al. (2014) measured the first natural frequency and the passive

deformation of a dragonfly wing. An artificial wing that exhibited the same characteristic

was fabricated and the first natural frequency was found to be 120 Hz. This proves that the

dragonfly wing does not flap at its natural frequency to avoid resonance effect. Previously,

researchers concluded that dragonfly wings flap at their natural frequency to conserve

energy (Seiichi et al. 2008). The wing beat frequency of a flapping dragonfly is 30 Hz.

Univers

ity of

Mala

ya

16

There is a twist at the nodus of the natural wing which increasingly deforms from the base

to tip. A carbon rod was used to replicate the leading edge. By this, the artificial wing was

able to produce a natural frequency similar to an actual dragonfly wing. The aerodynamic

power created was five times higher in magnitude than a carbon wing. Tanaka (2010)

demonstrated a method to fabricate corrugated artificial insect wings using a hoverfly as a

model. The wing, made of a thermosetting resin, contained veins on a corrugated

membrane. This mimics the wing structure of an actual hoverfly wing. The veins and

membranes are created by a single 3D mold. The molds were created using a layered UV-

laser micro machine.

The surface profile of the fabricated artificial hoverfly wing matched the original

mold. The spanwise stiffness of the artificial wing was 2.2 × 10−7 Nm−2, which is the

same magnitude of a natural hoverfly wing. This research has managed to produce an

artificial 3D insect wing with micron order precision. Shang (2009) fabricated an artificial

wing with defined static properties that were comparable to a natural wing (first

approximation). The method described by Shang can be modified to accommodate camber

or corrugated profiles. The polydimethylsiloxane (PDMS) mold can be shaped according to

the wing pattern and then cured. Spin-coating can be used to allow free flow of the mold

onto the venation pattern (although other methods are allowed as well). PDMS or another

appropriate polymer can be chosen as a membrane. Etching will further aid in the

distribution in a confined area. This can also aid in introducing complex folds and flexion

lines just like a biological wing. Dynamic bending and stiffness will be greatly affected by

wing mass disproportionality. Since the venation pattern correlates with mass distribution,

it is likely that biologically inspired artificial wings can be optimized to exhibit the

dynamic properties found in real dragonfly wings. Francis et al. (2001) have fabricated a

MEMS polymer structure which integrates the thorax, two wings, and a tergum. These

Univers

ity of

Mala

ya

17

parts were connected with a non-integrated actuator. Based on the experimental results and

observation, it was shown that this structure is able to adequately mimic the flexures and

twist of real wing motion.

2.6 Frequencies of insect wings

Several studies have been done to find the flapping and natural frequencies of flying

insects. Ngoc et al. (2013) conducted experiments to compare the natural frequencies of an

actual beetle hind-wing and an artificial wing which biomimics it. Both of these wings

were subjected to dynamic vibrations to measure resonance frequencies. The wing mass

was measured by an electronic balance. A total of 29 points on an actual beetle hind-wing

and 25 points on the fabricated artificial wing were painted to aid the laser sensor in

measuring the deflection of the wing. The recorded frequencies of beetle hind wing were

47.5 Hz, 88 Hz, and 176 Hz, respectively. Since the flapping frequency of a beetle ranges

from 35 to 40 Hz, it was assumed that the higher mode shape would have little effect on the

shape of the wing (deformation). Additionally, Ngoc et al. 2013, studied the relationship of

eight different insect species. The wingbeat frequency of these flying insects was measured

using a high-speed camera (imaging technique) while the natural frequency was determined

using a laser displacement sensor. Ngoc found that there is a prominent ratio between

wingbeat frequency and natural frequency. This ratio is 0.12-0.67 for insects flapping less

than 100 Hz and 1.22 for those with higher wingbeat frequencies. These findings suggest

that these frequencies do correlate with each other.

2.7 Static test conducted on real and artificial insect wings

Univers

ity of

Mala

ya

18

Rajabi et al. (2011) performed a detailed study on dragonfly wings using SEM.

These images are used to study the morphology and microstructure of an actual dragonfly

wing. A micro tensile tester was used to study the mechanical behaviors of the wing. An

overall observation shows that the wings have a highly corrugated structure. This is

believed to increase the overall stiffness of the wing and possibly prevent initial crack and

tear. This indirectly provides a tough barrier to the wing (preventing crack propagation by

closing the crack tip). The critical region of the wing is the nodus. The corrugations also

provide a high load bearing capacity and flexibility. The hollow shaped tubules can be

assumed to provide a high fatigue resistance especially when the wings are subjected to

dynamic loadings (flapping motion). They also increase the flexibility of the wing

structure, thereby affecting its fatigue resistance. Rajabi presents the fracture strains

obtained as a percentage value whereby the forewing and hindwing values are 5.65% and

5.58%, respectively.

Song et al. (2004) tested longitudinal veins (costa and radius) extracted from four

damselflies and tested in a micro-tensile machine. The specimens were held with a special

paper frame. The specimens were classified into two categories: fresh and dry. The average

tensile strength of the actual (fresh) and dry costa are 251 MPa, and 232 MPa. The average

modulus value for the actual (fresh) and dry costa are 14.37 GPa, and 14.43 GPa,

respectively. These values are found to be essentially the same which shows that aging

does not give any significant impact on tensile properties (costa was left to dry for a year).

The average tensile strength of fresh (actual) and dry radius are 285 MPa and 278 MPa.

The modulus of an actual (fresh) radius and dry radius are 16.87 GPa, and 14.89 GPa,

respectively.

2.8 Flapping mechanism

Univers

ity of

Mala

ya

19

Francis et al (2001) proposes a novel micro flying machine concept based on

MEMS technologies, which could potentially perform ascending, linear movement and

rotation. The principal advantage of using such technologies is that the solutions proposed

in the design process can be easily scaled down. The first step involves studying the

feasibility of design and machining a type of stainless steel artificial wing. Few flying

insects (weighing 1 gram) are able to flap their wings which have dimensions in the range

of 10mm to 30 mm. These insects flap with a frequency of 20 Hz to 150Hz. Secondly, a

linear electromagnetic actuator was proposed to excite the wings. Usage of the actuator has

proven advantageous as the excitation frequency can be adapted to the wing resonance

frequency. In the near future as proposed by the authors, a simple control mechanism can

be integrated to control the mechanism by having different oscillations on both sides (right

and left) without modifying the original mechanism. There should be more research done

focusing on the improvisation of the flapping mechanism.

Univers

ity of

Mala

ya

20

CHAPTER 3: RESEARCH METHODOLOGY

3.1 Introduction

This chapter introduces the methods employed to conduct the experiments and

simulations in this research project. It is ordered as follows: wing model overview,

topological optimization, method and types of analysis used, wing frame fabrication

process, static stress experiments, the flapping mechanism and the high speed imaging

technique set up.

3.2 Wing model overview

The dragonfly species used in this study is called Diplacodes Bipunctata. This

species has a distinct red discoloration with a wing span and body length of approximately

0.055 m and 0.035 m, respectively. A digital image (resolution 1600 dpi) was taken of this

dragonfly using a Nikon D7000 camera (Figure 3.1). Both the fore and hind wings were

modeled from this digital image.

Univers

ity of

Mala

ya

21

Figure. 3.1 Main parts of a dragonfly wing

a) Radial ; b) Costa- leading edge (rectangular pattern throughout) the stiffest pattern of all

parts; c) Nodus (part of the vein which is much thicker than the others, plays a pivotal role

in the dissipation and storage of mechanical energy); d) Pterostigma (behaves as a

dampener);

e) Anal margin and supercosta; f) Small hexagonal patterns which are compactly arranged

ranging from 0.0001 to 0.0002 m in size; g) Large quadrilateral and hexagonal patterns

ranging from 0.0002 to 0.0004 m in size. 1-15) Dragonfly wing patterns that are divided

into several regions for further study.

Univers

ity of

Mala

ya

22

3.3 Spatial network analysis

Spatial network analysis as mentioned in Chapter 2 is a graphical topological

optimization method that is used in various fields involved with geographical maps. A

spatial network (sometimes also called geometric graph) is a graph in which the vertices or

edges define spatial elements associated with geometric objects, which means the nodes are

located in a space within a specified unit (e.g. radius or distance). There are various

methods of conducting spatial network analysis. Geographical Information Systems used

this method to explore geographic spatial patterns (Angel, 2010). Our methodology is

similar, but we are applying this algorithm (based on shape, angle and density) to a

biological structure (a dragonfly wing). The idea of simplifying a dragonfly wing based on

spatial network analysis was inspired by observing its compactly arranged geometrical

patterns. Although there are a variety of methods available to conduct a spatial network

analysis, the proximity proposition method was chosen for this analysis because of its

conformity of use to a wing structure. The proximity proposition states that among all

shapes available, the circle has the shortest average distance to its center (Angel, 2010).

Since there are many possible centers for a known shape, it is essential to define at least

one appropriate center of a shape before defining a Proximity Index: The Proximate Centre

is defined as the center of gravity of a shape. The Proximity Index: ratio of the average

distance from all points in the equal-area circle center: average distance to the Proximate

Centre from all points in the shape.

Frolov (1975) computed that the mean distance to the centre of a circle is equal to two

thirds of its radius and suggested that it would be quite accurate to be used as a measure of

compactness. The Proximity Index therefore uses Frolov’s equation. In our case, this new

approach is initiated due to the fact that after close observations, it could be noted that the

Univers

ity of

Mala

ya

23

3.3.1 Canny edge detection algorithm

The image was imported into Matlab and segmented using the Canny edge

detection algorithm, resulting in a logical image illustrated with point clouds. This

grayscale image was then smoothed with a Gaussian filter to suppress noise. The main

vein structures were manually traced out and divided into sections using splines and

polygons. An edge detection algorithm was required to define the different regions

separated by veins in the image of the wing. The Canny edge detection algorithm is one of

the most popular algorithms, because of its detection of edge at a low error rate. The

algorithm runs in five steps which are:

1) Using Gaussian filter to smooth the image and remove noise;

All camera digital images will possess some amount of noise (blurry lines or edges,

dots, etc). Noise must be reduced to avoid errors caused by wrongly assuming the noise as

a part of the edge. This is an essential first step for Canny edge detection algorithm. The

image must then be smoothed using a Gaussian filter. The filter is applied using a matrix

quadrant as shown in (1) below:

1

𝑥 ||

𝑎 𝑏 𝑐 𝑏 𝑎𝑏 𝑑 𝑒 𝑑 𝑏𝑐 𝑒 𝑓 𝑒 𝑐𝑏 𝑑 𝑒 𝑑 𝑏𝑎 𝑏 𝑐 𝑏 𝑎

|| (3.1)

2) Edges are detected where the computed intensity gradients are the largest;

The Canny algorithm finds edges where the grayscale intensity of the image

changes the most. These areas are found by determining pixel gradients in the image.

Sobel-operator (feature detection filter in Matlab) is applied to determine the pixel

gradients in the smoothed image. Gradients are approximated in the x- and y-directions by

applying the kernels shown in (3.2) and (3.3). In these equations 𝐾𝐺𝑋 and 𝐾𝐺𝑌 represent

the kernel

Univers

ity of

Mala

ya

24

matrices and a and b are unknowns;

𝐾𝐺𝑋

= |−𝑎 0 𝑎−𝑏 0 𝑏−𝑎 0 𝑎

| (3.2)

𝐾𝐺𝑌 = |𝑎 𝑏 𝑎0 0 0

−𝑎 −𝑏 −𝑎| (3.3)

The strongest edges can then be assumed as an Euclidean distance. This can be measured

by applying the law of Pythagoras. It can also be simplified by applying Manhattan

distance formula as shown in Equation 3.2. The Euclidean distance value is then applied to

the image. The computed edge stresss are then compared to the smoothed image;

|𝐺| = √𝐺𝑥2 + 𝐺𝑦

2 (3.4)

where Gx and Gy are the gradients in the x- and y-directions respectively.

Equation 3.5 shows the method to determine the direction of the edges.

𝜃 = arctan (|𝐺𝑦|

|𝐺𝑥|) (3.5)

3) Local maxima are marked as edges;

The blurred edges are converted to sharp edges in this step. The highest value (maximum)

gradient is preserved and the others are deleted. The algorithm for each pixel in the gradient

image is outlined below:

1. Convert or round all the nearest θ (gradient) to the nearest 45◦.

2. The edge stress of the current pixel in both positive and negative directions are

compared. For example, if the gradient direction is north (90◦), compare with the pixels to

the north and south.

Univers

ity of

Mala

ya

25

3. Check to see that the edge stress of the current pixel is the largest. If not, remove the

value.

4) Use double threshold to detect potential edges;

The non-maximum values of the gradients may be the true edges of the image but

they were not strong due to the large effect of noise or color variation. The simplest way to

distinguish them is to use a threshold. The Canny edge detection algorithm uses a double

thresholding method (Matlab)

5) Track edges via hysterisis whereby final edges are determined by suppressing all non-

connected edges to a prominent edge.

Strong edges can immediately be included in the final edge image. Even if there is a

lot of noise interference, this noise will not be strong enough to mask strong edges. (for

threshold levels that are properly adjusted). Weak edges are included if they are deemed to

be a part of the true image and only if they are connected to strong edges. BLOB-analysis

(Binary Large Object) was used for edge tracking. This preserves strong edges are

preserved while suppressing weak edges.

3.3.2 Proximity index

The smaller vein structures were then modeled, in a similar way, to match the

detailed patterns and densities shown in the logical image. The spatial network analysis

method (proximity index) was employed to perform the segmentations. Although there are

a variety of methods available to conduct a spatial nework analysis, the proximity

proposition method was chosen for this analysis because of its conformity of use to a wing

structure. A circle is the most compact geometric shape possible, so this was selected as a

reference defining the specified unit (radius) (de Smith et al. 2007). The proximity

Univers

ity of

Mala

ya

26

proposition is defined as the shortest average distance of a circle in a given area of

symmetrical shapes. The proximity index is the ratio of the proximate center from all points

in the shape. Where (xi, yi) are the coordinates of a particular node (i), n is the total number

of nodes, and (xc, yc) are the fixed coordinates of bref (point of reference).

The mean distance (dCA) of any point bounded by the circle (of radius RA) to its center is

given in Eq. 7( de Smith et al. 2007):

𝑑𝐶𝐴 =1

𝜋𝑅2 ∫ 2𝜋𝑟2𝑅𝐴

0𝑑𝑟 =

2

3𝑅𝐴 (3.7)

Since

𝑅𝐴 = √𝐴 𝜋⁄ (3.8)

Where A is the area of the proximity circle

Therefore:

𝑑𝐶𝐴 =2√(𝐴 𝜋⁄ )

3 (3.9)

The proximity index (Iy) is the ratio of the proximate center from all points in the shape.

The 𝑑𝐶𝑆 is the radius of the circle. The equation for calculating the proximity index can be

written as:

𝐼𝑦 =𝑑𝐶𝐴

𝑑𝐶𝑆=

2√(𝐴 𝜋⁄ )

3𝑑𝐶𝑆 (3.10)

Nodes were manually placed at the vein structure intersections on the dragonfly wing

model selected for biomimicry (Figure 3.2). The nodes were only placed on intersections of

veins that are 3.0 x 10-7

μm thick or lesser. Each circle was manually sized to be the

minimum radius required to encompass all of the minor vein patterns (thickness less than

Univers

ity of

Mala

ya

27

3.0 x 10-7

μm) surrounding a single node. In this manner, all of the minor vein patterns

were encompassed in one of the many proximity circles inserted.

A fixed wing reference point (bref) is defined as the midpoint of the maximum spanwise

length of a single wing (from root to tip). The averaged distance to the fixed reference point

(bref) from all nodes is defined by (de Smith et al. 2007):

Figure 3.2: Spatial Network Model

[Note: nodes are widely spaced in an exaggerated manner for figure clarity]

Univers

ity of

Mala

ya

28

3.3.3 CAD model

An initial wire frame, two-dimensional scaled model was constructed using a CAD

software Rhinoceros 5.0. This was then exported into AutoCad 2015 to create a three

dimensional solid model (Figure 3.3 (a) and (b)). Although there is a difference in thickness

across the wing, it is small. For example, the vein at the base and the tip of the wing differ

by only 2 × 10−6

m (Li et al, 2009). Therefore in order to simplify the model, average vein

and membrane thicknesses of 2.5 × 10−6

m and 3.0 × 10−7

m were used, respectively

(Maria, 2011, S. Combes 2010, Jerzy, 2008).

(a)

(b)

Figure 3.3: (a) Digital image of the forewing of a dragonfly; b) Forewing created by Canny

edge algorithm; c) Canny edge forewing created after noise minimization and main pattern

identification; d) Forewing CAD model

Univers

ity of

Mala

ya

29

(c)

(d)

Figure 3.3: (continued) a) digital image of the forewing of a dragonfly; b) forewing

created by Canny edge algorithm; c) Canny edge forewing created after noise minimization

and main pattern identification; d) forewing CAD model

3.4 Finite element analysis (FEA)

FEA or more commonly known finite element analysis modeling follows three

general rules: pre-processing, analysis and post processing. Equation (3.11) is a general

structural equation (Bucalem, 1997):

|M| Ü + |C|u + |K|u= F cos (At + r) (3.11)

Univers

ity of

Mala

ya

30

Where |M| is the mass matrix, Ü is the acceleration, C is the damping matrix, K is the

stiffness matrix and F is the force vector. The software shall assemble the matrix equation

of the structure. The first part would be to solve the matrix equation below:

|M| Ü + |K|u = 0 (3.12)

This equation was used to solve the free vibrations of structure. The solution to the

equation above gives the natural frequencies (eigenvalues) and the undamped mode shapes

(eigenvectors). These are the parameters needed to define the basic dynamical properties of

a structure. Since, the main focus is natural frequency, which is a dynamic response, these

values are needed in subsequent analysis for dynamic displacements and stresses. For

harmonic motions:

U ̈= -ω^2 u (3.13)

In this equation ω is circular frequency. Substituting equation (3.14) into equation (3.13)

gives the matrix eigenvalue expression of:

|K|-1 = |M| u = |I| u (3.14)

In this equation |I| is the identity matrix. The back substitution method is used to obtain the

corresponding matrix of eigenvectors |Ҩ|. The matrix equation of motion for the structure

contains off-diagonal terms. The matrix equation may be deduced by introducing the

following transformation:

u = |Ҩ| x (3.15)

and writing the following expressions:

Univers

ity of

Mala

ya

31

|Ҩ|T |M||Ҩ| = |M*| (3.16)

|Ҩ|T |K||Ҩ| = |K*| (3.17)

|Ҩ|T |C||Ҩ| = |C*| (3.18)

In these equations Ҩ is modal displacement matrix, M* is modal mass matrix, K* is modal

stiffness matrix and C* is modal damping matrix. The expressions contain only diagonal

terms. The damping matrix may be uncoupled on the condition that the damping terms are

proportional to either the corresponding stiffness matrix terms or the corresponding mass

matrix terms. The uncoupled matrix expressions are:

|M*|x + |C*|x + |K*|x = |Ҩ|T F cos (At + r) (3.19)

Each equation in this expression then has the form:

Mrxr + Crxr + krxr =drcos (At + rr) (3.20)

to which the solution is:

𝑛𝑟 = 𝑑𝑟 cos(𝐴𝑡+𝑥𝑟)

√𝑘𝑟−𝑚𝑟𝑋2)2

+(𝑐𝑟𝑋)2 (3.21)

where c is modal damping coefficient, k is modal stiffness, and m is modal mass. The

dynamic displacements, u at frequency ω may then be obtained from the transformation.

u=|Ҩ|ŋ (3.22)

where ŋ is modal displacement.

Univers

ity of

Mala

ya

32

3.4.1 Element and mesh

The purpose of building an Autodesk model was to calculate the natural frequency

and displacement of the wing under a load. The mesh sizes of the elements were viewed

visually to ensure the mesh distribution was fine throughout the structure. The mesh was

chosen based on the results of a grid study. A grid study was performed to determine the

mesh size required for accurate results. The program was run with both coarse and fine

grids, with mesh sizes of 0.001 m and 0.0001 m respectively. The difference in results

between the coarse mesh and a fine mesh was found to be less than 2%. However, a

comparison of the run times of these mesh sizes was 70 % faster for the detailed model and

50% faster for the simplified models. (4 hours for detailed and 2 hours for simplified in the

former mesh size). Therefore a mesh size of 0.001 m was used for all subsequent runs.

Shell based elements are often used in finite element analysis models to calculate

the displacement. Each element in this model was specified as a shell, in order to simulate

the highly efficient load bearing capabilities of an insect wing (Jiyu et al, 2012, Combes,

2010, Tu Delft, 2013, Du Mont, 2009). Teflon was specified as the material for each shell.

This is because past research has identified Teflon as possessing similar characteristics to a

real dragonfly wing (made up of chitin and chitosan) (Antonia et al, 1998). The model is

assumed to have identical material properties in all directions with a single modulus of

elasticity and Poisson Ratio value. The properties are summarized in Table 3.1.

Table 3.1: Properties of Teflon

Density 1260 kg/m3

Modulus of elasticity 6.0 GN/m2

Poisson’s ratio 0.25

Univers

ity of

Mala

ya

33

Autodesk’s Algebraic Iterative Multi-Grid Scheme, with a third order Newton

Raphson integration method, was used as the finite element solver. A general element

formulation was selected, because this provides a robust solution for thin and thick

elements. This was necessary because the wing model is designed with two different type

of thickness (vein and membrane). The analysis formulation is set to nonlinear for natural

frequency analysis and linear for static stress analysis. The shell element model is set to be

isotropic, as it is assumed that the parts will only experience deflection in the elastic region

of the material.

3.4.2 Modal analysis: Mode shapes and MAC

A fixed constraint was placed at the base of the wing. This mimics a real dragonfly

which flaps with a fixed joint at the base. The modal analysis computes the natural

frequencies, mode shapes, the |M|, |C| and the response analysis matrices. Ten modes of

interest were computed in this experiment. The cut-off frequency was set to 100Hz as a

constraint for the simulation results. A lumped mass representation is used with an

allocated ratio of 0.2 percentage of data storage (S. Combes, 2003).

Correlation analysis of the similar mode shapes between the detailed and simplified

model were carried out. MAC (Modal assurance criterion) requires mode shape data from

two similar modes of the structure. Due to the large size of the finite element models in

terms of their number of nodes, it is not practically possible to include all the nodes into

analysis. Therefore 30 nodes were chosen, along the center of the wing span, at an interval

of 0.0002 m (distance between each node) were chosen as the representative of the mode

shape. The modal assurance criterion was calculated based on the mode shapes obtained

from the simulation results which will be discussed in Chapter 3. Mode was calculated

Univers

ity of

Mala

ya

34

using MathCAD Prime Express 3.0 (Parametric Technology Corporation, 2013) based on

this formula:

𝑀𝐴𝐶𝑛 =(𝑈𝑛

𝑇.𝐷𝑛𝑇)2

(𝑈𝑛𝑇.𝑈𝑛).(𝐷𝑛

𝑇.𝐷𝑛) (3.23)

where Un stands for detailed mode shape matrix of the nth

similar mode, Dn stands for the

simplified mode shape matrix of the nth

similar mode, UnT and Dn

T is the transpose of the

detailed and simplified mode shape matrices respectively and MACn is the value for the nth

similar mode shape (the dots represent multiplying these variables).

3.5 Artificial wing frames and nano-composite chitosan membrane

Several geometrically identical wings were fabricated based on the simplified

model created. The frames were fabricated from seven different materials: stainless steel

(Type 321), balsa wood, black graphite carbon fiber laminates, red prepreg fiberglass

laminates, polylactic acid (PLA), acrylic sheet and acrylonitrile butadiene-styrene (ABS).

Three samples of each wing type were tested to determine their tensile stress and bending

performance. The same nano-composite was used as a membrane for each wing frame

(which will be discussed in section 3.5.8). The mechanical properties of these materials are

as stated in Table 3.2. The compromise between minimizing the weight and maintaining

adequate tensile and bending stresss are critical for BMAV wings.

Univers

ity of

Mala

ya

35

Table 3.2: Mechanical properties of frame structure materials (David 2014, D. Akin et al

2014, Arcy, 1961, Shyy, 2010, Rahman, 2012, Marcin, 2011, Boeing, 2003 and 2004,

Borrega, 2015, Da Silva, 2007, Leban, 2013, and Osei, 2014)

Material Density

(kg/m3)

Modulus of

Elasticity

(N/m2)

Poisson

Ratio

Shear

Modulus of

Elasticity

(N/m2)

Thickness

(m)

Stainless steel (Type

321)

7920.0 1.90 x1011

0.3 7.7 x 1010

0.01

Balsa wood 130.0 3.00 x109

0.5 1.5 x 108

0.02

Black graphite carbon

fiber

1750.0 2.00x1011

0.5 2.4 x 1011

4x10-4

Red prepreg fiberglass 1522.4 2.00x1011

0.5 3.0 x 1010

3x10-4

Polylactic acid (PLA) 1190.0 3.50x109 0.36 3.37x10

9 2x10

-4

Acrylic 1180.0 3.32x109 0.35 6.20x10

7 2x10

-4

ABS 1050.0 2.80x109 0.35 1.03x10

9 2x10

-4

3.5.1 Fabrication of stainless steel (Type 321) wing frames

(a) (b)

Figure 3.4: Stainless steel type 321 wing frames; a) forewing, b) hindwing

Stainless steel (Type 321) is in the high carbon grade category, meaning that it

contains a minimum and maximum of 0.04% and 0.10% carbon, respectively. Higher

carbon helps the material retain stress at extreme temperatures. Type 321 is commonly used

in aircraft applications; because it is low in density (e.g. relative light weight) and high

temperature resistance. In our experiment, the stainless steel was subjected to solution

annealing. This is a common process done to ensure the smoothness of the grain structure.

This means only the carbides which may have precipitated to the grain boundaries are

Univers

ity of

Mala

ya

36

dispersed into the metal matrix by the annealing process. Type 321 stainless steel offer

higher creep and stress rupture properties.

The stainless steel wing frames were laser cut from sheet metal. These frames were

then coated with an anti-rust layer and spray painted using an acrylic lacquer spray. This

was necessary because the nano-composite film membrane will expose the wing frame to

oxidation. This is because the film contains concentrated glacial acetic acid and tannic acid

as a cross-linker. The anti-rust layer and acrylic lacquer spray hinder the oxidation due to

the presence of the acids. Oxidation is visibly indicated by purple blotches appearing on the

frame. After these processes, the frames were then immersed in the nano-composite

suspension which formed a film membrane coating the entire wing, as previously

described. Figure 3.4 shows the finished stainless steel wings.

3.5.2 Fabrication of balsa wood wing frames

Balsa wood is the most widely-used material used in hobby remote-controlled

aircraft. This is because of its extremely low density (making it very lightweight) and high

mechanical stress. Balsa wood was chosen for this study because it represents a minimum

weight solution. The balsa wood wing frames were cut manually with wood carving knives

and then immersed into the nano-composite solution. Figure 3.5 shows the finished balsa

wood wings.

(a) (b)

Figure 3.5: Balsa wood wing frames; a) forewing, b) hindwing

Univers

ity of

Mala

ya

37

3.5.3 Fabrication of black graphite carbon fiber wing frames

Black graphite carbon fiber sheets were purchased and visually inspected to make

certain they were uniform in appearance and free from foreign material. The specifications

are listed in Table 3.3.

Table 3.3: Specifications of black graphite carbon fiber

The woven laminate was hardened using an epoxy resin with a hardener (Araldite

rapid kit). Both epoxy resin and hardener were mixed with a ratio of 1:1 and then applied

evenly to the laminate in a thin layer. The laminate was then left to cure in a Memmert

UNB 300 convection oven for about two hours at a constant temperature of 45°C. The

forewing and hindwing shapes were then carved out of the hardened laminates manually

with a knife, using the stainless steel wings as a reference. The only difference between the

black graphite wings and the others, were that the inner gap regions were not carved out

according to the simplified model. Attempts to do so proved impossible, because it resulted

Thickness 0.04 mm

Carbon yarn spacing Four per 0.45 kilograms per strand

Warp tracer weave spacing 0.3 ± 1 m across the fabric width

Fill tracer weave spacing 0.6 ± 1 m apart

Length to diameter ratio (L/D) (See

Figure 3.6)

> 10

Figure 3.6: Waviness of carbon fiber yarns

Univers

ity of

Mala

ya

38

in delamination of the material. However, the overall size and shape of these wings are the

same as the others. Figure 3.7 shows the finished black graphite carbon fiber wings.

3.5.4 Fabrication of red prepreg fiberglass wing frames

Red prepreg fiberglass plies with a thickness of 0.03mm were purchased and

visually inspected. After curing, the red prepreg fiberglass exhibits a plastic-like

characteristic unlike carbon fiber which will retain its strand-like properties. This increases

the chance of delamination if a resin and hardener is not applied. The red prepreg fiberglass

(after curing) does not require an extra application of resin or hardener. In order to create

the wing frame structure, the material was warmed as necessary to enable easier manual

carving with a knife. Figure 3.8 shows the finished red prepreg fiberglass wings.

(a) (b)

Figure 3.8: Red pre-preg fiberglass wing models after immersion in chitosan nano-

composite solutions; a) forewing, b) hindwing

(a) (b)

Figure 3.7: Black graphite carbon fiber wing frames; a) forewing, b) hindwing

Univers

ity of

Mala

ya

39

3.5.5. Fabrication of polylactic acid (PLA) wing frames

The PLA wings were fabricated using a Maker Bot Replicator 2X 3D printer. PLA

is chosen for its biodegradability, lightweight, flexibility and elasticity (Joong, 2013). The

fabricated PLA wing models are shown in Figure 3.9.

(a) (b)

Figure 3.9: PLA wing models after immersion in chitosan nano-composite solutions; a)

forewing, b) hindwing

3.5.6 Fabrication of acrylic wing frames

The acrylic wings were fabricated using micro laser machining. Acrylic or

polyacrylate are generally known for their resistance to breakage, elasticity and flexibility

(David, 2014). Fabricated acrylic wing models are as shown in Figure 3.10.

(a) (b)

Figure 3.10: Acrylic wing models after immersion in chitosan nano-composite solutions;

a) forewing, b) hindwing

Univers

ity of

Mala

ya

40

3.5.7 Fabrication of acrylonitrile butadiene styrene (ABS) wing frames

Models made out of ABS are constructed from a thermoplastic. ABS is very useful

for functional applications because it matches 80% of the properties of the real injected

production material. ABS models are very accurate and have intermediate level of printed

details. You have a lot of freedom for the design of your model. However, the surface

quality of the models is rougher compared to other materials. The ABS wings were

fabricated using a Maker Bot Replicator 2X 3D printer. ABS is chosen due to its stress,

flexibility, and machinability (Hwan, 2012). The fabricated ABS wings are shown in

Figure 3.11.

(a) (b)

Figure 3.11: ABS wing models after immersion in chitosan nano-composite solutions; a)

forewing, b) hindwing

3.5.8 Chitosan nano-composite solution (membrane)

The chitosan nano-composite suspension as mentioned earlier was made up of a

chitosan suspension reinforced with nano-sized chitin whiskers and cross-linked using

tannic acid. This nano-composite film was processed by our research team for this specific

purpose and is featured in another article (Rubentheren et al, 2015). The suspension was

transformed into a thin 3 mm film by the casting evaporation method. This film was chosen

because it closely mimics the material properties of a dragonfly wing membrane. The

Univers

ity of

Mala

ya

41

mechanical properties and water resistivity of chitosan film can be controlled by the

addition of chitin whiskers and tannic acid as a cross-linker. Utilizing nano-sized whiskers

as a filler material elevates the composite film’s mechanical property (rigidity). A

crosslinking agent is also added to alter the mechanical properties of the film created.

Tannic acid is fully biodegradable and less expensive to produce, compared to the other

chemical derivatives. Tannic acid also has a high antioxidant capacity and can interact with

other biological macro-molecules. Addition of tannic acid as a cross-linking agent reduces

the swelling behavior, solubility and the rigidity of the nano-composite film. In order to

ensure that the immersion is uniform, the structure was submerged in the solution. This also

ensures that both sides of the frame structure are evenly coated with the solution. Once

cured, the film created a shiny, transparent film layer that adheres firmly to the frame

structure.

(a) (b)

Figure 3.12: Wing structure immersed in chitosan nanocomposite solution; (a) fore

wing, (b) hind wing

Univers

ity of

Mala

ya

42

3.6 Numerical Bend-Twist Coupling and Static Stress Simulation Analysis

3.6.1 Numerical bend-twist coupling analysis

The bend-twist coupling is one of the most relevant analyses for a wing cross

section. The effect of the bend-twist coupling appears in a wing structure as a link between

its two deformation states – bending and torsion. A static bend-twist coupling analysis was

performed for further validation of the models created. As previously described, the

analysis was set up with a constraint fixed at the base to mimic an actual dragonfly flapping

mode. Figure 3.13 shows the placement of constraints for a natural frequency analysis

3.6.2 Static stress simulation analysis

A fixed constraint was placed at the base of the wing tip to mimic the hinge of an

actual dragonfly wing. Nodal forces were applied at the tip of the wing to mimic the

tensile and bending test experiments. Simulations were done for all eight wings (forewing

and hindwing fabricated from all four materials). In addition, simulated bending and

tensile tests were done for the frame structures without the membrane to record the effects

of the chitosan nano-composite film on the frame structure. The stresses computed in the

static analysis are used to form the stress-dependent contribution to the tangent-stiffness

Figure 3.13: Placement of constraints fixed in a natural frequency analysis

Univers

ity of

Mala

ya

43

matrix. Static stress analysis, enables the study of stress, strain, displacement, and shear

and axial forces that result from static loading. Figure 3.14 shows the static stress

simulation model for a static stress analysis. Numerical analysis was not constructed until

failure due to the fact that dragonfly wings do not flap until they fail. Hence, a linear

model was used.

Figure 3.14: Constraints for a static stress analysis

3.7 Experimental Set Up

Tensile and bending experimental tests were done on the physical models using the

Instron Universal Testing Machine (Model 5569). Increasing loads were incrementally

applied to the wings until they failed. The data was compiled and integrated by Instron

MERLIN software. A specific jig was designed for both bending and tensile test to clamp

the fabricated models, whereby the slots were used to clamp the thin structures firmly. All

testing was repeated three times for each wing type, necessitating the fabrication of three

identical copies of each wing type. Since the wing frame test specimen is a sheet-like

structure, the same ratio of width-to-gage length (18:18 for forewing and 20:20 for

hindwing) was maintained in order to compensate for the elongation that occurs during

diffuse necking. ASTM test specifications were followed throughout the experiment. Each

base of the wing models were clamped at a fixed point and the tips were clamped at the

Univers

ity of

Mala

ya

44

moving point. The load was then increased gradually with a speed rate of 0.013 m/s. The

rate of extension and the stress-strain curve was observed until the wing model experienced

a structural failure. This point can be observed in the collected data as a sudden decrease in

the load. Data obtained in RAW file (not processed) was then converted to Microsoft Excel

for further analysis. The high speed camera imaging technique was then employed after

close observation of all these seven materials, and three best materials were chosen based

on performances. The main criteria of selection were the ability to withstand high tensile

stress and possess a high compression stress vs strain. Due to the complexity of an actual

dragonfly wing (due to the corrugations and veins) it was impossible to conduct tensile

testing on an actual wing. However, previous research has provided sufficient values to aid

in setting a benchmark for the criteria of the suitable material. Referring to previous

research, the value of tensile and compression stress for an actual dragonfly wing was

found to be 285 MPa and 16.87 GPa, respectively (Rupan et al, 2013). As mentioned

earlier, in order to test the aero-elastic properties of the dragonfly wings, the final step of

the research was two test it on an actual flapping mechanism. This is a static mechanism

which is able to produce a flapping motion frequency up to 250 Hz.

Previous researchers have conducted studies using high speed camera. Gui et al.

(2010) applied simple optical modifications to replace an expensive two-camera system for

taking stereo videos of fire ant alate wingbeat motion. Many have utilized the high speed

camera technique to capture wing motion of birds and insect flights. (Norris 2001; Sudo et

al. 2001, 2005; Dalton 2002; Gorman 2003). Recent advances in high-speed imaging

technique make it possible to capture insect wing motion not only at high phase resolution

but also at high digital resolution, e.g. the Photron Ultima APX high-speed camera used in

our lab is capable of capturing partial frames of 1024256 pixels at 8,000 frames per second

(fps). High digital resolution and high phase resolution together enable a precise analysis of

Univers

ity of

Mala

ya

45

the wing motion within a wingbeat period. However, at least two images from different

view angles, e.g. provided by a stereo imaging system, are necessary to determine the three

dimensional shape and orientation of the wing surface. A stereo imaging system requires at

least two cameras that view an object from two different directions, and in an ideal

situation, these two view directions should be perpendicular to each other, (Sudo et al.

2001, 2005).

Figure 3.15: Schematic diagram of Gui et al (2010) of fire ant alate wing motion

In order to capture the motion of the wing motion of the wings, two Phantom Miro

310 (Vision Research) high frame rate cameras were used to view the flapping wings from

two different directions in Figure 3.15. The camera’s high frame rate enables a precise

sequence of images to be captured of the flapping wing motion within a single wing beat.

Two cameras were necessary in order to determine the three-dimensional shape and

orientation of the wing surface (Figure 3.16 (a) and (b)). The cameras were placed

perpendicular to one another following the procedures established by Gui et al. (2010).

Univers

ity of

Mala

ya

46

Both cameras were equipped with a Nikon F lens. A multiple LED lighting system was

used to provide sufficient illumination. Imagery was recorded at a resolution of 320 x 240

pixels and a frame rate of 35000 per second, which allowed the wing beat motion to be

precisely captured. The motion video was stored in a computer via two high speed Ethernet

cables. It was played-back and analyzed using the Vision Research Phantom Camera

Control Software (version 2.6.749.0).

Univers

ity of

Mala

ya

47

(a)

Figure 3.16: (a) Actual experimental set up of the high speed camera imaging

technique, (b) Schematic diagram of the experimental set up

(b)

Figure 3.16: (continued) (a) Actual experimental set up of the high speed camera

imaging technique, (b) schematic diagram of the experimental set up

X 10

Univers

ity of

Mala

ya

48

(a) (b)

Figure 3.17: Front view (a) and side view (b) of the wing motion captured and

measurement axes.

Measurements were taken using each of the three wings while flapping at varying

frequencies: 10, 20, 30, 70, 120, 200 and 250 Hz. Figure 3.17 shows the front and side

view of the wing motions that were measured and recorded from captured imagery. Figure

3.17(a) illustrates the bending angle (θ) and displaced distance or deflection (d). Figure

3.17 (b) defines the wing tip angle (α) and the wing tip rotational twist speed (ω).

3.8 Flapping mechanism

The wing flapping mechanism used in this study is an electromagnetic flapping

wing actuator. The power supply used in this flapping wing drive is 9 volts DC. A LM555

crystal clock oscillator integrated circuit (shown in Figure 3.18) is used to generate a stable

oscillation. The free running frequency and duty cycle are accurately controlled with two

resistors and one capacitor. The generated oscillation is fed to a Power MOSFET fast

switch. The output of the Power MOSFET is used to actuate the miniature PC Board Relay.

The frequency of the switch (corresponding to the wing beat frequency) can be adjusted by

a 22 kW potentiometer. Each of the different wings is attached to a flat iron plate (2 mm

long and 2.75 mm thick) using super glue. This plate (wing platform) is oscillated by an

electromagnetic actuator (3 x 3 mm). Figure 3.18 shows the wing structures attached to the

actuator. The plates are attached to the hinge of the wing to mimic the joint of an actual

Univers

ity of

Mala

ya

49

dragonfly. This flapping mechanism is able to create a linear up-down stroke motion at

variable wing beat frequencies, up to a maximum frequency of 250 Hz. The flapping

degree was set to be 60° which corresponds with an actual dragonfly wing flapping angle

during hovering flight (Jane, 2004 and Xinyan, 2008).

Figure 3.18: Flapping mechanism used in this research

Univers

ity of

Mala

ya

50

CHAPTER 4: RESULTS AND DISCUSSION

4.1 Introduction

This chapter includes discussion pertaining to the results obtained. It includes the

simulation results, the experimental results, application of the chosen wings in a flapping

mechanism and the images of this application.

4.2 Simplified model creation

Two models were created for each wing. The first model (or detailed model shown

in Figure 3.3) closely approximates the fine detailed structure of an actual dragonfly wing.

This was built using the Canny edge detection algorithm previously described in Section

3.3.1. Due to its small size and complexity, this model proved to be too difficult to fabricate

using laser micro machining. Therefore a second model (or simplified model) was created

that simplifies the structure into its major constituents, by grouping dense patterns and vein

structures into single regions. Fabricated simplified model examples are shown (Figure

4.1). These samples were laser cut from thin (1 mm thickness) stainless steel sheet metal.

Careful consideration was taken into keeping within machining tolerance, with the

objective of fabricating a wing for a BMAV design in the near future. The ratio of the

quadrilateral, pentagons and hexagons are 10:30:60 based on observation. Therefore,

although the hindwing has a larger area than the forewing, the number of regions created is

justified by considering the ratio of patterns. Examples of dragonfly wing patterns can be

seen in Figures 4.2 (a) and (b). The costal part of the wing consists of several uniformally-

arranged rectangular membrane patterns. These were grouped together in different regions

(consisting of 10 to 15 rectangles per region). There are many small, hexagonal membrane

patterns at the anal margin and wing tip. These were grouped into several regions

(consisting of 30 to 50 patterns per region). Table 4.1 shows a summary comparing the

pattern densities of the detailed and simplified models.

Univers

ity of

Mala

ya

51

(a)

(b)

Figure 4.1: Fabricated simplified model (a) Forewing; (b) Hindwing

Table 4.1: Model specifications in terms of number of membranes, density and patterns

The spatial network analysis method was implemented by dividing the wing model

into circles with diameters of 0.00025m and 0.0005 m, for smaller and larger regions

respectively. The size of regions correspond to the pattern density in that particular area.

Smaller regions are used in higher density areas. The center of the circle was determined by

looking at the starting point of a branch of the main vein. Previous research was used as a

reference to determine the center of the circle (HuaiHui et al. 2013).

CAD model Properties

No. membrane

regions

Area ( m2

) Average pattern

density

(region/m2)

Forewing Hindwing Forewing Hindwing Forewing Hindwing

Detailed

Model

787 900 0.00055 0.000577 14309 15517

Simplified

Model

17 16 0.00055 0.000577 30900 27586

Univers

ity of

Mala

ya

52

(a)

(b)

Figure 4.2: Typical dragonfly wing membrane pattern arrangement

Figures 4.2 (a) and (b) shows membrane patterns that are typical of dragonfly

wings. A number of proximate cells are located between two main veins and arranged in

one row to form quadrilateral shapes. Sometimes there are two rows of cells, intercalated

between a pair of main veins. As shown in Figure. 4.2 (b) in this case, one row will fit into

another at an angle of 120° as a result of coequal tension, but both cells are at right angles

when it meets the main vein as in the former case. For numerous rows of cells, all the

angles between them tend to be co-equal angles of 120°, and hence the cells resolve into a

hexagonal or polygonal meshwork. The cell arrangement in dragonfly wings was explained

through soap bubble experiment (Arcy,1961). When four soap bubbles meet in a plane,

they arrange in symmetrical ways either with four or five partition-walls intersecting at

right angles. Five partition walls are known to be unstable. The point where the patterns

branch into four to five patterns was determined to be the center of the circle. The hindwing

of a dragonfly is slightly broader in size compared to the forewing, but the pattern it

comprises remains the same (e.g. quadrilateral, pentagons and hexagons). The number of

patterns that make a region is therefore drawn by observation with the appropriate ratio

(10:30:60) applied to it.

Figure 3.2 in the previous section, shows how the circles were placed on the

detailed model in order to formulate what will become the simplified model. The nodes

(that determine the center of each proximity circle) were placed at branching points,

according to the soap bubble theory. A past article reports that a good index value should

Univers

ity of

Mala

ya

53

be close to 1.0 (de Smith, 2007). This value indicates the compactness of the geometric

pattern enclosed by the circle. If the index is within an acceptable range, this validates the

creation of a polygonal region within the selected area. The proximity index for the circles

were found (from Equation 10) to be between 0.66 to 0.83.

It is difficult to fit circles into regions near edges of the wing that are highly

cambered (Figure 4.3). Trying to do so will either create large gaps or lap over the edge.

Because of this, point clouds were used instead of circles in these regions. The Canny edge

detection algorithm played a pivotal role in determining the shape of a region by

highlighting the vein with the strongest edge from the orientation of point clouds. The

shape of the bounded region was drawn by connecting these strong edges (Figure 4.3).

Figure 4.3: Example of connecting strong edges using cloud points

Figures 4.4 (a) and (b) shows the resulting simplified model of the forewing and

hindwing, respectively. The overall shape and size of the detailed and simplified model are

the same, only the internal vein and membrane pattern was redesigned.

Univers

ity of

Mala

ya

54

4.3 Mode shape analysis and MAC

The mode shape similarities are shown in Figures 4.5 and 4.6. Out of ten mode

shapes computed, the five best matching mode shapes were selected and compared between

these two models. The modes were distinguished by the deflection shape of the structure.

Higher mode shapes were included because these mode shapes will be able to determine

the flutter speed in future work. Therefore Figures 4.5 and 4.6 also include the higher mode

shape comparison. The natural modal frequencies were also listed in Figures 4.5 and 4.6. It

is known from the structural dynamics that natural frequency is directly proportional to the

square root of the stiffness and inversely proportional to square root of the mass. Due to the

minimization of patterns (reduction of number of regions), the overall mass has been

minimized with the detailed model having a mass of 0.0216 kg and the simplified model

having a mass of 0.0182 kg. This reduction has been achieved by utilizing spatial network

analysis. Minimization of mass and reduction in stiffness will result in an increase of

natural frequency and the results shows a good agreement with this. The increase in natural

frequency was in the range of 3-10% for all simplified models.

(a) (b)

Figure 4.4: Simplified models, (a) forewing, (b) hindwing

Univers

ity of

Mala

ya

55

f= 21.1 Hz f= 22.2 Hz

Mode 2 Mode 1

(a)

f = 39.9 Hz f = 41.9 Hz

Mode 3 Mode 2

(b)

f = 59.1 Hz f = 61.1 Hz

Mode 5 Mode 6

(c)

f = 68.3 Hz f = 73.7 Hz

Mode 6 Mode 5

(d)

f = 94.6 Hz f = 99.9 Hz

Mode 8 Mode 8

(e)

Figure 4.5:Mode shapes and corresponding natural frequencies for both detailed (left) and

simplified models (right) : forewing

Univers

ity of

Mala

ya

56

f = 10.0 Hz f = 11.4 Hz

Mode 2 Mode 1

(a)

f = 24.4 Hz f = 27.9 Hz

Mode 3 Mode 2

(b)

f = 38.2 Hz f = 40.6 Hz

Mode 4 Mode 3

(c)

f = 44.4 Hz f = 48.5 Hz

Mode 6 Mode 4

(d)

Figure 4.6: Similar mode shapes and corresponding natural frequencies for both detailed

(left) and simplified models (right) : hindwing

Univers

ity of

Mala

ya

57

The slight difference in natural frequency between the models is due to the fact that there

are fewer membranes in the simplified models, which reduces the torsional rigidity of the

wing (HuaiHui, 2013). The calculated natural frequency and deformed modals are in

accord with previous vibration test experiments that have been conducted (Ching, 2009,

Rajabi, 2011). The mode shapes of the wings are defined as: up-down flapping, bending,

and twisting (Rajabi, 2011). Figures 4.5 (a) and 4.6 (a), shows the detailed wing model

with a prominent up-down flapping mode; a bend upwards at the posterior main vein; and

a twist at the group of transverse vein at the wing base. Similarly, Figures 4.5 (c), (d) and

4.6 (c), (d) shows the detailed wing model with up-down flapping and a significant bending

and twisting along the anterior and posterior of the group of main veins. Figures 4.5 (e) and

4.6 (e) shows that the simplified models exhibit a slight up-down flapping with a clear bend

at the wing tip. This agrees with previous research conclusions that the membranes

increase wing component rigidity (Du Mont, 2009, Combes, 2003, Tanaka and Shyy,

2010).

The MAC values from Equation (23) obtained for the five compared mode shapes

are shown in Table 4.2. A MAC value indicates that a unique mode shape should ideally be

1 and it reduces to 0 with reducing degree of correlation.

f = 62.9 Hz f = 65.6 Hz

Mode 7 Mode 7

(e)

Figure 4.6: (continued) Similar mode shapes and corresponding natural frequencies for

both detailed (left) and simplified models (right) : hindwing

Univers

ity of

Mala

ya

58

Table 4.2: MAC values for all five compared mode shapes for both forewing and hindwing

nth mode

shape

1 2 3 4 5

Forewing 0.709 0.987 0.944 0.664 0.909

Hindwing 0.935 0.948 0.768 0.622 0.602

The values shown in Table 4.2 prove that both the detailed and simplified models have a

close degree of correlation which further validates the simplifed model (Rahman, 2012).

The range of accepted values would be from 0.65 - 0.9 (values which are closer to 1). It can

be noted that there is a slight difference in mode no 4 which is negligible (0.622) since the

percentage difference is 4.6%.

4.4 Numerical bend-twist coupling analysis

Anisotropic materials have elastic couplings related to shear and in‐plane/out

plane/out‐of‐plane behavior. Most often undesired behavior include warping. Some of the

elastic couplings include bending that induces twist. (bend-twist coupling). Elastic coupling

in structures is a direct effect of having different laminate modulus in different directions,

known as anisotropy. The bend-twist couplings in laminates are created by a variation in

thickness (plane modulus) which is not balanced by an opposing coupling effect of another

ply. Dragonfly wings are anisotropic in nature which means that there is a significant

amount of bend-twist effect in them. In order to be structurally correct the simplified wing

model has to possess the same equivalent axial and bending stiffness as the detailed

(original) wing in order to properly predict wing deflections and stresses. Ideally, the wing

tip deflection of the simplified model should match the deflection of the real wing, for any

given applied load.

By graphically comparing the detailed and simplified model plots (Figures 4.7 and

4.8) it could be observed that the twist angles for the simplified models are higher

Univers

ity of

Mala

ya

59

(approximately 3-5%) than those of the detailed model. Moreover, the trend of deflection

(bending) is the same in all plots confirming the observation. The basal leading edge of

dragonfly wings are highly corrugated structures, which improves its flexural and bending

stiffness (Combes, 2003 and Tanaka, 2010). This prevents chord-wise flattening especially

at the main veins. None of our models includes this complexity. The lack of corrugated

edges will generally not be an issue for a BMAV wing if rigid, metallic structures are used,

rather than the chitin-based structures in actual dragonfly wings. Studies have shown that

the corrugation has only a slight effect on the wing’s aerodynamics, because the scale of

corrugation is small compared to the separated flow region and size of the leading edge

vortex (Xue Li, 2009). Therefore, corrugated wings will provide structural benefits without

significant effect on aerodynamic lift production.

Univers

ity of

Mala

ya

60

Figure 4.7: The static bend-twist coupling graph of both detailed and simplified models;

forewing

Univers

ity of

Mala

ya

61

Figure 4.8: The static bend-twist coupling graph of both detailed and simplified models;

hindwing

Univers

ity of

Mala

ya

62

4.5 Tensile test simulation results

Simulation results were obtained before the experiments were conducted. Both

tensile and bending experiment results were analyzed and the stress distribution was

observed. Simulation tensile test results were observed and critical or high stress areas

were marked as points of fractures. These results helped in identifying the weak areas

(regions) of the wing models of different materials and the maximum stress that the models

could withstand.

4.5.1 Tensile test simulation results (wing frames without membrane)

The first experimental tests involved the wing frame structures without their

membranes. (The material specifications for these models are stated in Table 3.2.) Figure

4.9 shows the von Mises stress results of all the seven different frame structures. The ‘max’

(maximum) and ‘min’ (minimum) pointers indicate the area where the stress is the highest.

The highest stress in the forewing recorded for stainless steel (Type 321), balsa wood,

black graphite carbon fiber,red prepreg fiberglass, PLA, acrylic and ABS is: 2.54N/mm2,

0.04N/mm2, 2.03 N/mm

2, 0.03 N/mm

2, 13 N/mm

2, 17 N/mm

2, 23 N/mm

2 respectively. A

close observation from the results shows that the high stress point for all seven materials,

except for balsa wood and acrylic, occurs in the same region. This region is located where

the surface-to-area ratio is at its minimum value. The maximum stress on the balsa wood

wing occurred at the grip point. Based on previous studies (Da Silva, 2007), balsa wood,

and acrylic to a certain extent, under compression in the axial direction exhibits a linearly

elastic regime that terminates by the initiation of failure in the form of localized kinking.

Geometric imperfections in the form of fiber waviness, and failure was found to

lead to kink bands with distinct orientations and widths (Kyriakides, 1995). Subsequently,

Univers

ity of

Mala

ya

63

under displacement-controlled compression, a stress plateau is traced that is associated with

the gradual spreading of crushing of the cells through the material. The material is less stiff

(i.e. weaker) in the tangential and radial directions. Compression in these directions crushes

the tracheids (phloem of the wood) laterally. Figure 4.10 shows the results for the

hindwing. The maximum stress occurs in approximately the same location for all the wing

frame structures. This region corresponds with the minimum surface-to-area ratio. The

minimum stress also occurs in a similar location for all of the wing frames, except for the

balsa wood and acrylic frame. The readings are 2.31 N/mm2, 0.08 N/mm

2, 1.39 N/mm

2,

0.02 N/mm2, 12 N/mm

2, 17 N/mm

2 and 24 N/mm

2 for stainless steel (Type 321), balsa

wood, black graphite carbon fiber and red prepreg fiberglass, PLA, acrylic and ABS

respectively. The minimum stress for the balsa wood and acrylic frame occurs in the

cambered edge, because it experiences the lowest compression stress. This has been

observed for balsa wood in past studies (Borrega, 2015, Osei, 2013 and 2014)

(a) (b)

(c) (d)

Figure 4.9: Von Mises stress simulation results of different forewing frame structures; (a)

stainless steel (Type 321), (b) balsa wood, (c) black graphite carbon fiber, (d) red prepreg

fiberglass e) PLA, f) acrylic, g) ABS

Univers

ity of

Mala

ya

64

(e) (f)

(g)

Figure 4.9 (continued): Von Mises stress simulation results of different forewing frame

structures; (a) stainless steel (Type 321), (b) balsa wood, (c) black graphite carbon fiber, (d)

red prepreg fiberglass e) PLA, f) acrylic, g) ABS

(a) (b)

(c) (d)

Figure 4.10: Von Mises stress simulation results of different forewing frame structures; (a)

stainless steel (Type 321), (b) balsa wood, (c) black graphite carbon fiber, (d) red prepreg

fiberglass, e) PLA, f) acrylic, g) ABS

Univers

ity of

Mala

ya

65

(e) (f)

(g)

Figure 4.10 (continued): Von Mises stress simulation results of different forewing frame

structures; (a) stainless steel (Type 321), (b) balsa wood, (c) black graphite carbon fiber, (d)

red prepreg fiberglass, e) PLA, f) acrylic, g) ABS

Univers

ity of

Mala

ya

66

4.5.2 Tensile test simulation results (wing frames with membrane)

Simulation results were also obtained for the wing frames with their membranes.

Figure 4.11 shows the forewing models of all four materials. Based on Figure 4.11, the

maximum von Mises stress occurs at approximately the same location for all seven

materials. This location is different from the models without membranes (Figure 4.9).

However, the highest stresses occur again in regions where the surface-to-area ratio is

minimum. An exception is seen in the wing frame made of black graphite carbon fiber

laminate, where the maximum stress occurs in the cambered edge of the wing. This is due

to the fact that carbon fiber undergoes delamination at areas where the kinks and curves are

located. (Tamilarasan, 2015). The region where maximum stress occurs shows a potential

area of delamination. The maximum stress recorded is: 2.05 N/mm2, 0.097 N/mm

2, 49.3

N/mm2,

2.10 N/mm2, 14.77 N/mm

2, 17.29 N/mm

2, and 24.23 N/mm

2 for stainless steel

(Type 321), balsa wood, black graphite carbon fiber, red prepreg fiberglass, PLA, acrylic

and ABS, respectively.

Univers

ity of

Mala

ya

67

Figure 4.11: von Mises stress simulation results of different forewing model structures; (a)

stainless steel (type 321), (b) balsa wood, (c) black graphite carbon fiber, (d) red prepreg

fiberglass, (e) PLA, (f) acrylic, (g) ABS

Univers

ity of

Mala

ya

68

The stress distribution of the hindwings with their membranes (shown in Figure

4.12) is also different than the wings without their membranes (shown in Figure 4.10). The

maximum stresses for all of these wing models occurs on the membrane and not the frame.

This is due to compression of the membrane in that region. This indicates that there is a

high probability that the membrane in this region will be damaged. The maximum stress

recorded is: 3.28 N/mm2, 1.04 N/mm

2, 41.05 N/mm

2, 3.09 N/mm

2, 13.87 N/mm

2, 16.96

N/mm2, 23.46 N/mm

2 for stainless steel (Type 321), balsa wood, black graphite carbon.

(a) (b)

(c) (d)

(e) (f)

Figure 4.12: von Mises stress simulation results of different hindwing model structures; (a)

stainless steel (Type 321),(b) balsa wood, (c) black graphite carbon fiber, (d) red prepreg

fiberglass, (e) PLA, (f) acrylic, (g) ABS

Univers

ity of

Mala

ya

69

(g)

Figure 4.12: (continued) von Mises stress simulation results of different hindwing model

structures; (a) stainless steel (Type 321),(b) balsa wood, (c) black graphite carbon fiber, (d) red

prepreg fiberglass, (e) PLA, (f) acrylic, (g) ABS

4.5.3 Bending test simulation results (wing frames without membrane)

Some of the wing frames without membrane were subjected to bending tests. Only

the steel and balsa wood frame structures were subjected to this bending test. Attempts to

conduct bending tests on the red prepreg fiberglass, black graphite carbon fiber laminates,

PLA, acrylic and ABS wing frame structures gave trivial results. These materials do not fail

when subjected to bending test experiments (Babukiran et al, 2014). Therefore, the

simulation results for black graphite carbon fiber, red prepreg fiberglass laminates, PLA,

acrylic and ABS were not included.

Figures 4.13 and 4.14, shows the von Mises stress simulation results for the

forewing and hindwing structures. The bending simulation results show uniformity in the

stress distribution of all four frame structures for both forewing and hindwing. The

maximum stress occurred in the same relative region for all four different frame structures,

regardless of the wing type. This shows that bending creates a localized stress in both of the

materials used. Combes et al. 2003 found that the flexural stiffness across the wing base

line is maximum when it is subjected to bending (Combes, 2003). The hinge of a dragonfly

Univers

ity of

Mala

ya

70

wing must be able to withstand different types of loads, including forces subjected to

rotational and translational motion. Hence the maximum stress occurred at the region near

the wing base (where the pivot point is located), as highlighted in the simulation results.

The maximum bending stresses for the forewings of stainless steel and balsa wood are:

0.80 N/mm2 and 0.11 N/mm

2, respectively. The hindwing maximum stress was: 0.64

N/mm2 and 0.05 N/mm

2 for stainless steel and balsa wood, respectively.

(a) (b)

Figure 4.13: von Mises stress simulation results of different forewing model structures; (a)

stainless steel (Type 321), and (b) balsa wood

(a) (b)

Figure 4.14: von Mises stress simulation results of different hindwing model structures; (a)

stainless steel (Type 321), and (b) balsa wood

Univers

ity of

Mala

ya

71

4.5.4 Bending test simulation results (wing frames with membrane)

(a) (b)

Figure 4.15: von Mises stress simulation results of different forewing model structures; (a)

stainless steel (Type 321), and (b) balsa wood

(a) (b)

Figure 4.16. Von Mises stress simulation results of different hindwing model structures;

(a) stainless steel (Type 321), and (b) balsa wood

Figures 4.15 and 4.16 shows the bending stress created on the wing frames with

their membrane. The results show that the tips of the wings undergo damage faster than the

base of the wing. The figures also indicate that there is a greater probability of the

membrane undergoing damage than the frame structure. The region where the tip of the

wing undergoes bending deformation matches previous studies that have been conducted

(HuaiHui, 2012 and 2013, Wakelling, 1997, Jiyu Sun, 2012 and Jongerius, 2010). The von

Mises stress for the hindwing is 2.09 and 0.09 N/mm2

for stainless steel and balsa wood,

Univers

ity of

Mala

ya

72

respectively. However, only the results obtained here clearly show the effects of the

bending load on the membrane of an artificial dragonfly wing.

4.5.5 Tensile test experimental results

Figures 4.17 and 4.18 compare the forewing and hindwing for all seven different

frame structure materials. They also compare the results of models with and without

membranes, allowing the effects of the chitosan film on the frame structures to be

observed. All four materials (without membrane) exhibit their original properties under

tensile stress, which agrees with previous studies (David, 2014, Borrega, 2015, da Silva,

2007, Osei 2013 and 2014, Wang, 2013, Warren, 2015, and Tamilarasan, 2015). Both

figures show a sudden decline in tensile stress, which indicates failure (fracture) has

occurred (destructive testing). Figure 4.17(a) compares the steel forewing with and without

membrane. The ultimate strength of the steel forewing with its membrane is 2.10 MPa and

without membrane is 2.48 MPa. These values are 2.20 MPa with membrane and 2.45 MPa

without membrane for the hindwing. These results show that the addition of the film

membrane causes an 8% reduction of strength for the steel wing. The chitin nano-

composite membrane solution oxidizes the stainless steel wing frame. We observed

aggressive oxidation in our samples, even though precautions were taken to avoid it.

Therefore, it can be concluded that adhesion of the film membrane to the stainless steel

reduces its overall stress resistance and increases its oxidation. Therefore, stainless steel is

a poor choice of material for the wing frame, when desiring to use a chitin nano-composite

film wing membrane. Referring to Figures 4.11(a) and 4.12(a), the percent difference

between the simulation and experiments are approximately 4.0% and 3.52%, with and

without membranes, respectively. The simulation results do not take into count the decrease

Univers

ity of

Mala

ya

73

of strength caused by chemical degradation. The percentage results shows a good

agreement between the experiment conducted and the simulation analysis obtained.

Figures 4.17(b) and 4.18(b) shows a comparison of balsa wood with and without

membrane for both fore and hindwings. In contrast to stainless steel, the membrane reduces

the stress acting on the balsa wood wing frame. The ultimate strength of the balsa wood

forewing, with and without membrane, is 0.09 MPa and 0.03 MPa, respectively. The results

of the hindwing, with and without membrane, were: 0.09 MPa and 0.04 MPa, respectively.

This equates to an increase in ultimate strength that ranges from 125% to 200%. Therefore

it can be concluded that adhesion of the chitosan nano-composite film to the balsa wood

increases its strength tremendously. Referring to Figures 4.11(b) and 4.12(b), the percent

difference between the simulation and experiments are approximately 6.23% and 4.56%,

with and without membranes, respectively.

Figures 4.17(c) and 4.18(c) shows a comparison of carbon fiber wings, with and

without film membranes. The adhesion of chitosan nano-composite film yields an immense

increase in strength. The peak tensile strength for the forewings, with and without

membranes, are 48.07 MPa and 1.32 MPa, respectively. The results for the hindwings, with

and without membranes, are 40.06 MPa and 1.34 MPa, respectively. This shows that the

addition of film increases the ultimate strength by 3000% to 3600%. This shows that the

adhesion of the film to these carbon fiber models greatly intensifies its ultimate strength.

Referring to Figures 4.11(c) and 4.12(c), the percentage differences between the simulation

and experiments are approximately 5.23% and 3.87%, with and without membranes,

respectively.

Figures 4.17(d) and 4.18(d) shows the results obtained for red prepreg fiberglass

model wings. These results also show an increase in ultimate strength after adhesion of the

chitosan nano-composite film. The ultimate strength for the forewing, with and without

Univers

ity of

Mala

ya

74

membranes, are: 2.03 MPa and 0.01 MPa, respectively. The ultimate strength for the

hindwings, with and without membranes, is 3.04 MPa and 0.01 MPa, respectively.

However, the red fiberglass undergoes warping and shrinkage at a very early stage of the

experiment (1 minute after starting the experiment). So this phenomenon must be taken

under consideration when considering this material for fabrication purpose. Referring to

Figures 4.11(d) and 4.12(d), the percent differences between the simulation and

experiments are approximately 5.75% and 8.93%, with and without membranes,

respectively.

Figures 4.17(e) and 4.18(e) shows a comparison of PLA wings, with and without

film membranes. The adhesion of chitosan nano-composite film yields an increase in

strength. The peak tensile strength for the forewings, with and without membranes, is

200.07 MPa and 175.46 MPa, respectively. The results for the hindwings, with and without

membranes, are 197.33 MPa and 162.06 MPa, respectively. This shows that the addition of

film increases the ultimate strength by 14% to 22%. This shows that the adhesion of the

film to these PLA models moderately increases its ultimate strength. Referring to Figures

4.11(e) and 4.12(e), the percentage differences between the simulation and experiments are

approximately 2.62% and 3.81%, with and without membranes, respectively.

Figures 4.17(f) and 4.18(f) shows a comparison of acrylic wings, with and without

film membranes. The adhesion of chitosan nano-composite film yields noticeable amount

of increase in strength. The peak tensile strength for the forewings, with and without

membranes, is 173.27 MPa and 168.95 MPa, respectively. The results for the hindwings,

with and without membranes, are 188.71 MPa and 179.16 MPa, respectively. This shows

that the addition of film increases the ultimate strength by 2% to 5%. This shows that the

adhesion of the film to these acrylic models slightly increases its ultimate strength.

Referring to Figures 4.11(f) and 4.12(f), the percentage differences between the simulation

Univers

ity of

Mala

ya

75

and experiments are approximately 2.12% and 3.55%, with and without membranes,

respectively.

Figures 4.17(g) and 4.18(g) shows a comparison of ABS wings, with and without

film membranes. The adhesion of chitosan nano-composite film yields noticeable amount

of increase in strength. The peak tensile strength for the forewings, with and without

membranes, is 265.77 MPa and 275.23 MPa, respectively. The results for the hindwings,

with and without membranes, are 244.88 MPa and 257.61 MPa, respectively. This shows

that the addition of film increases the ultimate strength by approximately 5%. This shows

that the adhesion of the film to these acrylic models slightly increases its ultimate strength.

Referring to Figures 4.11(g) and 4.12(g), the percentage differences between the simulation

and experiments are approximately 1.73% and 1.56%, with and without membranes,

respectively.

Univers

ity of

Mala

ya

76

(a)

(b)

(c)

Figure 4.17: Tensile stress vs tensile strain (engineering) of all four forewing frame

structures ; a) stainless steel (Type 321), b) balsa wood, c) black graphite carbon fiber, d)

red prepreg fiberglass, e) PLA, f) acrylic and g) ABS

Univers

ity of

Mala

ya

77

(d)

(e)

(f)

Figure 4.17: (continued) Tensile stress vs tensile strain (engineering) of all four forewing

frame structures ; a) stainless steel (Type 321), b) balsa wood, c) black graphite carbon

fiber, d) red prepreg fiberglass, e) PLA, f) acrylic and g) ABS Univ

ersity

of M

alaya

78

(g)

Figure 4.17: (continued) Tensile stress vs tensile strain (engineering) of all four forewing

frame structures ; a) stainless steel (Type 321), b) balsa wood, c) black graphite carbon

fiber, d) red prepreg fiberglass, e) PLA, f) acrylic and g) ABS

(a)

(b)

Figure 4.18: Tensile stress vs tensile strain (engineering) of all four hindwing frame

structures; a) stainless steel (Type 321), b) balsa wood, c) black graphite carbon fiber,

d) red prepreg fiberglass

Univers

ity of

Mala

ya

79

(c)

(d)

(e)

Figure 4.18: (continued) Tensile stress vs tensile strain (engineering) of all four hindwing

frame structures; a) stainless steel (Type 321), b) balsa wood, c) black graphite carbon

fiber, d) red prepreg fiberglass

Univers

ity of

Mala

ya

80

(f)

(g)

Figure 4.18: (continued) Tensile stress vs tensile strain (engineering) of all four

hindwing frame structures; a) stainless steel (Type 321), b) balsa wood, c) black graphite

carbon fiber, d) red prepreg fiberglass

4.5.6 Bending test experimental results

Figures 4.19 and 4.20 show the compression stress vs strain of the wing models

made of stainless steel and balsa wood, respectively. Figures 4.17 and 4.18 showed a

reduction in the ability of the wing frame to undergo tension when the film membrane is

adhered to the stainless steel. However, this testing shows that its resistance to compression

is higher in the wing models with film. Figures 4.19(a) and 4.20(a) show results for the

stainless steel frames, both the fore and hindwings. The peak strength is 2.02 MPa with

membrane and 0.75 MPa without membrane. This is an increase of 167%. The hindwings

are 2.03 MPa and 0.63 MPa, with and without membrane, respectively. This is an increase

Univers

ity of

Mala

ya

81

of 140%. Referring to Figure 4.15 and 4.16, the percent difference between the simulation

and experiments are approximately 9.97% and 8.74%, with and without membranes,

respectively.

Figures 4.19(b) and 4.20(b) show results for the balsa wood frame structures. The

peak strength for the forewing structure, with and without membrane, is 0.08 MPa and 0.04

MPa, respectively. The hindwings are 0.08 MPa and 0.04 MPa, with and without

membrane, respectively. This is an increase of approximately 200% for both balsa wood

fore and hindwings. Referring to Figure 4.16, the percent difference between the simulation

and experiments are approximately 6.49% and 7.77%, with and without membranes,

respectively.

(a) (b)

Figure 4.19: Compression stress versus compression strain graphs of forewing

wing models. (a) stainless steel (Type 321), (b) balsa wood

Univers

ity of

Mala

ya

82

4.6 High speed camera image results analysis

The high speed camera imaging experiment was conducted on each of the three

types of wings; PLA, acrylic and ABS (both with and without the chitosan membranes).

This was done to study the flexibility of each wing frame material and to determine the best

material for use in a BMAV. An actual dragonfly wing (Diplacodes Bipunctata) was also

tested to study its motion during passive flapping at different frequencies and compare it

with the fabricated wings. The nomenclature for wing rotation about different axis is shown

in Figure 4.21. This figure shows the types of rotation that an insect wing undergoes.

Figures 4.22 and 4.23 shows a sequence of images, illustrating the wing motion of an actual

flapping dragonfly wing during one complete flapping cycle. The wing beat frequency for

these images is 30 Hz, which is the nominal wing beat frequency of this species of

dragonfly. These figures show the common mode of flapping of an actual insect wing

(dragonfly). At all frequencies from 10-250 Hz, the flapping motion is similar. As for the

fabricated wings (all seven materials), the models only exhibit an up-down flapping

motion. They lack the twist and bend motion exhibited in an actual insect wing.. The

Figure 4.20: Compression stress vs compression strain for hindwing models. (a)

Stainless steel (Type 321), (b) Balsa wood

Univers

ity of

Mala

ya

83

twisting motion of an actual dragonfly wing can be seen clearly in Figures 4.22 (b), (c), (d),

(e) and 4.23(b), (c), (d), and (e). This twisting motion cannot be achieved by our fabricated

wing models. This will affect the aerodynamics performance of the wing. However, the

wing exhibits shown by these (PLA, acrylic and ABS) wings. Based on our previous

simulation and experimental analysis mentioned in Section 4.5 and 4.6, it has been

concluded that out of the seven materials, the three best materials were PLA, acrylic and

ABS. These materials were chosen for following reasons:

The thickness of the material was much thinner compared to steel, balsa,

carbon fiber and red prepreg ( less than 2mm). This matches the criteria of a

dragonfly wing whereby the wings are supposed to be thin.

These materials were much more flexible than the first four materials

mentioned above (steel, balsa, carbon fiber and red prepreg)

The simulation and experimental results shows a better agreement of when

compared to an actual dragonfly wing. The percentage difference of strength

between the actual dragonfly wing ( based on previous literature) and the

three materials (PLA, acrylic and ABS) ranged about 2%-5% which further

strengthens the rationale of selecting these materials

Furthermore, in the flapping mechanism experiment, only the forewings were chosen to be

tested. The difference between the forewing and hindwings tested previously (as mentioned

in Section 4.5 and 4.6) yields strength results that were similar (with a percentage

difference of 1%-2%) for both forewing and hindwing (with and without membrane).

Also, according to previous literature, the hindwings act as a wing that generates lift during

the flapping motion (Zheng Hu, 2009) Our experiments were based on a non-vacuum

Univers

ity of

Mala

ya

84

environment, (open air), hence the larger surface area of the hindwing shall affect the

measurement parameters greatly. (bend angle, displaced distance and twist angle)

Figure 4.21: Degrees of freedom for the wings of flying insects

Univers

ity of

Mala

ya

85

a)

b)

c)

d)

e)

f)

Figure 4.22: Side view of the actual dragonfly flapping wing (gray scale) captured by the

high speed camera during one flapping cycle at 30Hz. a) start of downstroke, b) mid-

downstroke, c) end of downstroke, d) start of upstroke, e) mid-upstroke, f) end of upstroke

Univers

ity of

Mala

ya

86

a)

b)

c)

d)

e)

f)

Figure 4.23: Front view of the actual dragonfly flapping wing captured by the high speed

camera (gray scale) during one flapping cycle at 30 Hz. a) start of downstroke, b) mid-

downstroke, c) end of downstroke, d) start of upstroke, e) mid-upstroke, f) end of upstroke

Univers

ity of

Mala

ya

87

Dragonfly wings greatly deform during flight. This was observed in our experiment

as well as others (Hisayoshi, 2014). Despite having a certain degree of rigidity, dragonfly

wings undergo a considerable amount of bending, twisting and rotational motions. Figure

4.22 and 4.23 shows the motion of flapping wing in one complete cycle at 30 Hz (side and

front view). It was observed that at both directions (chord and spanwise) an asymmetric

twist-bend motion was observed. Figures 4.22(d), 4.22 (f), and 4.23(d) clearly show these

asymmetric motions mentioned. At the end of an upstroke (observed in Figure 4.22(e), the

wing momentarily exhibited a symmetrical twisting motion. A large feathering rotation

range of 154° to 179° of the entire wing was observed during the beginning of the down

stroke and end of the upstroke (for all frequencies) (Figure 4.23(a) and 4.23(e). Even

during the steady phase (passive moment occurring when the flapping angle is zero), the

wing is observed to undergo internal torsion. This corresponds well to previous studies

made by Wootton et al. (Wooton, 1981 and 1993)

Besides the nominal 30 Hz wing beat frequency, the dragonfly wing was also

flapped at frequencies of: 10, 20, 70, 120, 200, and 250 Hz. The pattern of deformations

was similar for all of the frequencies observed. The measured bending angle, wing tip

deflection, wing tip twist angle and speed for the different wing frames (without and with a

membrane) were plotted for comparison to the results obtained from an actual dragonfly

wing in Figures 4.24-4.27.

4.6.1 Bending angle versus flapping frequency

The bending angle is directly proportional to the flexibility of the wing. Previous

studies have shown that wing flexibility has a significant influence on the aerodynamic

forces (e.g. lift and thrust) generated by the flapping wing ( Zhao, 2009). However, this

study focuses only on the chordwise flexibility of a passive flapping wing. Bending angles

Univers

ity of

Mala

ya

88

were measured along the chordwise direction. Chang et al.(2013) also investigated

chordwise flexibility, but for simple, non-anisotropic wing structures. They presented a

detailed assessment of the effects of structural flexibility on the aerodynamic performance

of flapping wings. The Reynolds number (Re =100) considered in this study is relevant to

small insect flyers, such as fruit flies. However, this study only includes the role of

chordwise flexibility and passive pitch in two dimensional plunging motions.

Our study involves a much more complex wing design than this past study by

Chang, 2013.. Figure 4.24 shows the bending angles as the wing beat frequency is varied

for the three fabricated wing frames (without and with a membrane) in comparison to an

actual dragonfly wing. These figures show that the maximum bending angle (θmax) for all

the wings occurs during the upstroke. This was observed for both frames (without and with

a membrane). This agrees with previous research done by Jongerius et al. (2010) in which

this asymmetry (difference in bending angle between the upstroke and down stroke) was

attributed to the directional bending stiffness in the wing structure (e.g. one-way hinge or a

pre-existing camber in the wing surface)

The maximum bending angle of dragonfly wings at 30 Hz is recorded to be about

6°. The wings were observed to have a maximum bending angle of 10.7° at 120 Hz (natural

frequency of an actual dragonfly). This is an increase of 78.3% from 30 Hz. ABS shows a

high level of flexibility compared to the other two materials used. Figure 4.24 shows that

the bending angle curves of the fabricated ABS wings are more similar to the actual

dragonfly wing than the other two types. Figure 4.24(a) shows that the bending angle of

ABS wing (without membrane) at 30 Hz is 8.5° and 5.9° at 120 Hz. At 30 Hz, the

percentage difference between an ABS wing (without membrane) and an actual dragonfly

wing is about 41.7%. The PLA and acrylic wings each recording a percentage difference

(reduction) of 70%. In Figure 4.24(b), ABS exhibits much larger bending angles at 30 Hz

Univers

ity of

Mala

ya

89

when the membrane is added. The value of the ABS wing (with membrane) is 20.1° at 30

Hz and 34.9° at 120 Hz. This angle is much larger than the actual dragonfly wing. The

percentage increase between the ABS and an actual dragonfly wing is 233.3%. The other

two materials (PLA and acrylic) exhibited much lower bending angles than the actual

dragonfly wing. The percentage reduction in PLA and acrylic (in comparison to an actual

dragonfly wing) is 83.3% and 75%, respectively.

These observations confirm that the overall flexibility of the wing decreases after

the membrane is attached, except for ABS. At a frequency of 120 to 170 Hz, the dragonfly

wing bends at a very high angle. Previous research shows that dragonflies do not flap at

their natural frequency (120 to 170 Hz) (Maria, 2011). So this result is likely due to a

resonance effect caused by the wing beat frequency being proximate to the natural

frequency of the wing. This result confirms that dragonflies have a maximum wing beat

frequency limitation in this range. The ABS wing frame shows a similar trend at 120 Hz.

The bending angle is reduced at frequencies greater than 120 Hz for both the actual

dragonfly wing and the three fabricated wings.

(a) (b)

Figure 4.24: Bending angle of different wing frames; (a) without membrane and (b) with

membrane.

Univers

ity of

Mala

ya

90

4.6.2 Wing tip deflection versus flapping frequency

Figures 4.25(a) and (b) show the wing tip deflection for varying wing beat

frequencies of the three fabricated wing frames (without and with membranes) in

comparison to an actual dragonfly wing. Similar to bending angle, deflection is another

measurement that can be used to assess a flapping wing’s flexibility. As mentioned earlier,

past studies have shown that wing flexibility has a significant effect on the wing’s ability to

generate a suitable time-averaged lift or thrust (Zhao, 2009). Similar to θmax in Figure 4.24,

Figure 4.25 shows that the maximum deflection (dmax) occurs during the upstroke. This

again was observed for both frames (without and with a membrane). This agrees with

previous research done by Luo et al (2012).

Figure 4.25 (a) shows that all of the fabricated wing frames (without membrane)

deflect at magnitudes that are similar (only slightly reduced) to the actual dragonfly wing at

30 Hz which is about 7.1 mm. At 30 Hz, ABS has a percentage increase of 23.94%. PLA

and acrylic both have a percentage reduction of 47.71% and 62% respectively. However

Figure 4.25 (b) shows that the fabricated wing frames (with membranes) have very

different deflections than the actual dragonfly wing. Only the ABS wing showed a

comparable level of deflection, however the dragonfly wing is 40.85% lower than the ABS

wing. The PLA and acrylic wings have percentage reduction of 94.37% and 66.2%,

respectively compared to the dragonfly wing. The actual dragonfly wing is able to undergo

a large deflection at the tip region. This supports previous studies which explain that the

difference between the deflection at the tip and the surface is created by differences of the

rigidity (due to the vein and corrugations) along the wing surfaces (Charles, 2011).

The difference in deflection between wing frames without a membrane and with a

membrane shows that the attachment of a membrane causes an increase in rigidity. This

increase in rigidity is observed to be the highest in the PLA wing. Only the ABS wing

Univers

ity of

Mala

ya

91

shows a similar curvature trend with the actual dragonfly wing around 120 Hz. At 120 Hz,

an increase in percentage of 81.7% (without membrane) and decrease in percentage of

69.7% (with membrane) is seen in ABS wing frame. Compared to the PLA wing there is a

percentage reduction of 82.6% (without membrane) and 64.2% (with membrane). The

acrylic wing has a percentage reduction of 85.3%, both without and with the membrane

attached. The trend of the graph again shows that there is a decrease in flexibility after the

membrane has been attached. Two high peaks were observed for an actual dragonfly wing

(30 and 120 Hz). As already stated, the natural frequency of dragonfly wings has been

reported to be between 120 to 170 Hz (Maria, 2011). The extreme fluctuation observed in

this range confirms the observation.

(a) (b)

Figure 4.25: Wing tip deflection of different wing frames; (a) without membrane and (b)

with membrane

4.6.3 Wing twist angle versus flapping frequency

Figure 4.26 shows the maximum wing tip twist angle of the three fabricated wing

frames in comparison to an actual dragonfly wing. The maximum twist angle was recorded

during the stroke reversal (transition from upstroke to down stroke). The twist angle for an

actual dragonfly wing at 30 Hz is 154.58°. Untwisted wings have large; drag producing

Univers

ity of

Mala

ya

92

wing surfaces that are exposed to flow hence, the importance of twisting in wings is

justified. Wing tip twist also plays an important role in enhancement of flight performance.

The mid-stroke timing of wing deformation in the butterfly, examined by Lingxiao et al

(2013), suggests that the deformation is not due to wing inertia, because the acceleration of

the wing is small at this point in the stroke. They suggest that this is instead due to

aeroelastic effects, since the aerodynamic forces are very large at mid-stroke.

Figures 4.26 (a) and (b) show that both the PLA and acrylic wing frames (both

without and with membranes) closely match the performance of an actual dragonfly wing.

At 30 Hz, the ABS (without and with membrane) has a percentage reduction of 19.8% and

1.10% respectively in comparison to the actual dragonfly wing. The PLA wing (without

and with membrane) has a percentage increase of 5.2% and 9.7% respectively. The acrylic

wing (without and with membrane) has a percentage increase of 7.1% and 11.7%

respectively. At 120 Hz, the ABS and acrylic wings (without membrane) has a percentage

reduction of 10.2% and 2.5%, respectively compared to the dragonfly wing. The PLA wing

(without membrane) has a percentage increase of 2.9%. The ABS wing (with membrane)

has a percentage reduction of 35.9% compared to the dragonfly wing. While the PLA and

acrylic wings have a percentage increase of 5.3% and 3.6% respectively. Based on these

results, the PLA and acrylic wings are more similar to the actual dragonfly wing than the

ABS wing. The large fluctuation of the ABS wing across varying flapping frequencies (10

to 250 Hz) makes it a more complicated BMAV option. Another trend observed from

Figure 4.26 is that the wing tip twist angle of the dragonfly wing does not vary significantly

as the flapping frequency is varied. This matches a previous study by Zhao et al. (2009).

(mentioned earlier) which shows that the flexibility of insect wings increases more

chordwise than spanwise, due to the rigid leading edge vein. This is true for both categories

of wing frames (with and without membrane).

Univers

ity of

Mala

ya

93

(a) (b)

Figure 4.26: Wing twist angle of different frames versus flapping frequency; (a) without

membrane and (b) with membrane

4.6.4 Wing tip twist speed versus flapping frequency

Figure 4.27 shows the wing tip twist speed for the three wing frames (without and

with membranes) in comparison to an actual dragonfly wing. Vogel (2013) stated that the

wing tip twist speed varies according to size and must exceed a ratio with flight speed

(wing tip twist speed: flight speed) by 3.7 or more to enable forward flight. Figure 4.27

shows that the PLA and acrylic wing frames (both without and with membranes) show a

similar curvature trend with the actual dragonfly wing. The wing tip twist speed of an

actual dragonfly wing at 30 Hz is 9.2 revolutions per second. At 30 Hz, the PLA wing

shows a percentage increase of 33.3% (without membrane) and percentage reduction of

52.2% (with membrane) in comparison with the dragonfly wing. The acrylic wing shows a

percentage increase of 30.4% (without membrane) and 44.4% (with membrane). The ABS

wing shows a percentage reduction of 67.4% (without membrane) and 64.1% (with

membrane). At 120 Hz, all of the fabricated wing frames without the membrane attached

show a slight percentage increase in comparison to an actual dragonfly wing. The ABS,

Univers

ity of

Mala

ya

94

PLA, and acrylic wings show a percentage increase of 6.4%, 4% and 5%, respectively.

While the ABS, PLA and acrylic wing frames without membrane have a percentage of

37.5%, 35.14% and 37.4%, respectively. The ABS wing frame shows a much different

curvature trend than the others, both with and without membrane. Figure 4.27 shows that

the wing tip twist speed is highly dependent on the flapping frequency and is less

influenced by changes in the frame’s flexibility. This can be confirmed by observing the

curves of the wing frames with membrane. The observed trend is the same across varying

flapping frequencies (10-250 Hz) for both types of wing frames.

(a) (b)

Figure 4.27: Wing tip twist speed; a) without membrane and b) with membrane

As mentioned before, the study conducted by Yang et al (2012), concludes that the

spanwise flexible deformation should be limited to a small range to achieve higher

aerodynamic performance for a flapping MAV. Alternatively a larger chordwise

deformation could serve to enhance the aerodynamic performance (e.g. lift and thrust

generation). The results of our experiments in flapping an actual dragonfly wing support

this observation, by showing that chordwise deformation is very significant (Figures 4.24-

Univers

ity of

Mala

ya

95

4.27) compared to the spanwise deformation. These results suggest that BMAV wings

should be designed with a stiff leading edge to limit the spanwise deformation and flexible

ribs to keep chordwise deformation within a significant but suitable range. This indicates

that the ABS wing design is better suited for use in a BMAV than the PLA and acrylic

wing designs. The dragonfly wing was used as a comparison to test the effectiveness of the

fabricated wings. No attempts were made to remove the membrane of an actual dragonfly

wing since the idea was to test all the fabricated materials (with and without membrane) in

comparison to an actual wing. From the graphs above, it could be noted that the fabricated

wings were much more rigid in certain ranges of frequency (significantly from 200-250

Hz). These results show that the fabricated wings were not able to match the higher degree

of flexibility exhibited by a dragonfly wing at larger frequencies. However, in the case of

an actual dragonfly as stated earlier, the wing beat frequency is 30 Hz and the natural

flapping frequency is 120 Hz. Hence, the rigidity of the fabricated wings beyond these

frequencies is of least concern.

Univers

ity of

Mala

ya

96

CHAPTER 5: CONCLUSION

5.1 Conclusions

The overall objective of this study was to produce a working simplified wing

model based on a dragonfly wing. In order to achieve our objective, various methods were

employed which resulted in our final product. Firstly, a methodology for producing

accurate but simplified dragonfly wing models was described. The first step was to create a

detailed model, using the Canny edge detection method on digital images of dragonfly fore

and hind wings. The spatial network analysis method was then used to create simplified

models from this detailed model. Both sets of models were used to calculate the natural

flapping frequency, deformation, and displacement using Autodesk. It was shown that the

simplified models produced very similar results (less than 13% difference for all

calculations). Therefore, this shows that the spatial network analysis is a suitable approach

to simplify a complicated insect wing structure (like a dragonfly). It is possible to fabricate

this simplified model, because it is within available micro laser cutting and 3D printing

allowances which enable its fabrication.

Secondly, a study that compares the mechanical properties of fourteen wing

structures and analyzes the effect of adhering a chitosan nano-composite film membrane.

Both FEA simulations and experiments were performed and all of the results showed good

agreement (less than 10% difference). Each of the seven materials considered for the wing

frame exhibited different characteristics. These seven materials were chosen because of

their potential suitability for use as wing frames on a BMAV. Stainless steel (Type 321) has

a high load bearing capacity, but experimental tests with steel wing frames showed that the

addition of the film reduces its tensile resistance and causes aggressive oxidation. Steel is

also relatively heavy making it a poor choice for this application. In contrast, film adhered

Univers

ity of

Mala

ya

97

to a balsa wood wing frame increased its tensile strength by 125% to 200%, while the peak

bending strength was improved by about 200%. However, both steel and balsa wood is

relatively inflexible compared to the other two materials. The black graphite carbon fiber

shows a remarkable Results show an increase in ultimate strength after it is adhered with

the chitosan nano-composite film membrane. The ultimate strength for the forewing, with

and without membranes are: 2.03 and 0.01 MPa, respectively. The ultimate strength for the

hindwings, with and without membranes, are 3.04 and 0.01 MPa, respectively. The primary

drawbacks in using red pre-impregnated fiberglass is that it undergoes warping and rapid

shrinkage. The load bearing capacity is also low compared to the other materials.

Use of PLA materials, fabricated using a 3D printer were also examined. The PLA

wings showed a remarkable load bearing capacity. The peak tensile strength for the

forewings, with and without membranes, is 200.07 MPa and 175.46 MPa, respectively. The

results for the hindwings, with and without membranes, are 197.33 MPa and 162.06 MPa,

respectively. This shows that the addition of film increases the ultimate strength by 14% to

22%. This shows that the adhesion of the film to these PLA models moderately increases

addition of the film reduces its tensile resistance and causes aggressive oxidation. Steel is

also relatively heavy making it a poor choice for this application. In contrast, film adhered

to a balsa wood wing frame increased its tensile strength by 125% to 200%, while the peak

bending strength was improved by about 200%. However, both steel and balsa wood are

relatively inflexible compared to the other two materials. The black graphite carbon fiber

shows a remarkable load bearing capacity and its lightweight property makes it suitable for

BMAV applications. Its primary disadvantage is the practical difficulties involved in

carving it into the simplified wing frame structure. Wing frames fabricated from the red

pre-impregnated fiberglass mimics the elasticity and flexibility of an actual dragonfly wing

structure, making it the most suitable material to be used. Results show an increase in

Univers

ity of

Mala

ya

98

ultimate strength after adhesion of the chitosan nano-composite film. The primary

drawbacks in using red pre-impregnated fiberglass is that it undergoes warping and rapid

shrinkage. The load bearing capacity is also low compared to the other materials. The PLA

wings showed a remarkable load bearing capacity. The acrylic wings, exhibited similar

results to PLA wings with the adhesion of chitosan nano-composite film significantly

increased the strength. The adhesion of chitosan nano-composite film to ABS wing models

yields noticeable amount of increase in strength. It shows that the adhesion of the film to

these acrylic models slightly increases its ultimate strength. The percentage differences

between the simulation and experiments are approximately 1.73% and 1.56%, with and

without membranes, respectively.

One of the many challenges faced in constructing a working BMAV, involves the

need to fabricate a highly deformable and flexible wing that has a large load bearing

capacity. The third phase of the research was to conduct an experimental study to assess

aero-elastic properties of flapping wings fabricated from the three chosen materials (PLA,

acrylic, and ABS). These materials were chosen due to a number of criteria; the thickness

of the material was much thinner compared to steel, balsa, carbon fiber and red prepreg

(less than 2mm). This matches the criteria of a dragonfly wing whereby the wings are

supposed to be thin. These materials were much more flexible than the first four materials

mentioned above (steel, balsa, carbon fiber and red prepreg). The simulation and

experimental shows a better agreement of mechanical strength when compared to an actual

dragonfly wing compared to the first four materials. The percentage difference between the

actual dragonfly wing (based on previous literature) and the three materials (PLA, acrylic

and ABS) ranged about 2%-5% which further strengthens the decision of selecting these

materials. The structural design of each of these wings is identical and based on

biomimicry of an actual dragonfly wing. The experimental results were compared to the

Univers

ity of

Mala

ya

99

actual dragonfly wing, on which they are based, in order to assess their potential

application to a BMAV design. A flapping mechanism that uses an electromagnetic

actuator is used. This mechanism was used to flap the wings at various frequencies from 10

to 250 Hz. A high frame rate imaging system, that uses two cameras, was used to capture

the three dimensional motion of the flapping wing. Several different aero-elastic parameters

were measured: bending angle, wing tip deflection, wing tip twisting angle, and wing tip

twisting speed. Analysis of wing bending angle and wing tip deflection indicates flexibility

of the wing in the chordwise direction, while the wing tip twist angle and speed shows the

flexibility of the wing in the spanwise direction. The ABS wing exhibited the highest

chordwise flexibility (indicated by their large bending angles and wing tip deflections).

Although the PLA and acrylic fabricated wings exhibited a much lower chordwise

flexibility than the ABS fabricated wing and the dragonfly wing, their spanwise flexibility

(indicated by their wing tip twist angles and speeds) closely matched the dragonfly wing.

These experimental results show that an actual dragonfly wing has a highly

deformable structure despite its rigidity. Our materials, though possess a certain amount of

flexibility, they were unable to match the twisting motions exhibited by an actual dragonfly

wing. The materials examined in this study (PLA, acrylic and ABS) were selected due to

their high flexibility, low density, and low fabrication costs. The ABS wing design gave

better results in matching the chordwise flexibility of the actual dragonfly wing, while

limiting the spanwise flexibility to much greater degree than the other two designs.

Univers

ity of

Mala

ya

100

5.2 Future Work

The choice of materials available to model a wing frame is limitless. There are

many other new emerging materials such as ultra-PLA, and nano-ceramics which can be

explored to fabricate these wing frames. The basic idea of creating a simplified model has

been given in this research. This model can be used for future studies. Despite its

corrugated structure and brittle nature, an actual dragonfly wing can exhibit a high degree

of flexibility. It is hoped that by exploring different wing frame materials, this degree of

flexibility can be achieved. The idea of the simplified model can further be expanded to

create a 3D model which encloses the tubules (hollow structures in the frame) if possible. It

is hoped that this optimization (simplification) method will be further explored and will

open up new possibilities to mimic various insect wings (for BMAV) application purposes.

Univers

ity of

Mala

ya

101

REFERENCES

Ada Silva and Kyriakides S. (2007) Compressive response and failure of balsa wood.

International Journal of Solids and Structures, 44, 8685-8717.

Aimy Wissa, Jared Grauer, Nelson Guerreiro, James Hubbard Jr, Cornelia Altenbuchner,

Yashwanth Tummala, Mary Frecker, Richard Roberts. (2015). Free Flight Testing

and Performance Evaluation of a Passively Morphing Ornithopter. International

Journal of Micro Air Vehicles. 7(1), 21-40.

Akın D, Kasgoz A, and Durmus A. (2014). Quantifying microstructure, electrical and

mechanical properties of carbon fiber and expanded graphite filled cyclic olefin

copolymer composites. Composites Part A: Applied Science and Manufacturing,

60(5), 44-51.

Akira Azuma, Tadaaki Watanabe. (1988). Flight Performance of a Dragonfly. Journal of

Experimental Biology. 137 (1) 221-252.

Angel S. J.P, Daniel L. Civco.(2010). Ten compactness properties of circles: measuring

shape in geography. The Canadian Geographer, 4, 441-461.

Antonia B Kesel U.P, Wener Nachtigall. (1998). Biomechanical aspects of the insect

wings: an analysis using the finite element method. Computers and Biology in

Medicine, 28, 423-437.

Babukiran B.V and Harish G. (2014). Influence of Resin and thickness of Laminate on

Flexural Properties of Laminated Composites. International Journal of Engineering

Science and Innovative Technology (IJESIT) 3.

Bao X.Q, Dargent T, Grondel S, Paquet J.B, and Cattan E. (2011). Improved

micromachining of all SU-8 3D structures for a biologically-inspired flying robot.

Microelectronic Engineering, 88 (8), 2218-2224.

Boeing. (2000). Boeing Material Specification - Glass fabric reinforcements for laminated

plastic products. Boeing

Univers

ity of

Mala

ya

102

Boeing. (2003). Boeing Material Specification - Glass fabric preimpregnated epoxy resin

low temperature curing. Boeing.

Boeing. (2004). Boeing Material Specification - Carbon fiber reinforcements, yarn and

fabric (BMS 8-9J). Boeing.

Borrega M and Gibson L.J. (2015). Mechanics of balsa (Ochroma pyramidale) wood.

Mechanics of Materials,84(5),75-90.

Bucalem M.L. (1997) Finite Element Analysis of Shell Structures. Archives of

Computational Methods in Engineering. 4, 3-61.

Butler, Rhett. (2015). Biomimetics, technology that mimics nature. Retrieved from

Mongabay website http://news.mongabay.com/2005/07/biomimetics-technology-

that-mimcs-nature/

Chang-kwon, Kang WS.(2013). Scaling law and enhancement of lift generation of an

insect-size hovering flexible wing. Journal of The Royal Society Interface 10:0361.

Charles Richter HL.(2011). Untethered hovering flapping flight of a 3D-printed mechanical

insect. Artificial Life.17.73-86.

Ching Hsin Chang and Kuen Tsan Chen. (2009). Study on Nanomechanical Properties of

Dragonfly Wing. Advanced Materials Research,79-82.1325-1328.

Combes S.A (2010). Encyclopedia of Aerospace Engineering. Materials, structure, and

dynamics of insect wings as bioinspiration for MAVs. Department of Organismic

and Evolutionary Biology, Harvard University, Cambridge, MA, USA: John Wiley

& Sons.

Combes S.A. (2003a). Flexural stiffness in insect wings I. Scaling and the influence of

wing venation. Journal of Experimental Biology, 206, 2979-2987.

Combes S.A. (2003b). Flexural stiffness in insect wings II. Spatial distribution and

dynamic wing bending. Journal of Experimental Biology, 206, 2989-2997.

Univers

ity of

Mala

ya

103

Combes S.A. (2003c). Into thin air: contributions of aerodynamic and inertial-elastic forces

to wing bending in the hawkmoth Manduca sexta. Journal of Experimental Biology,

206, 2999-3006.

Cox A., Monopoli D., Goldfarb M., Garcia E. (1999, September) Development of

piezoelectrically actuated micro-aerial vehicle. Paper presented at the SPIE

Symposium on Microrobotics and Microassembly. 3834, 101–108.

Dalton R. (2002) High speed biomechanics: Caught on camera. Nature. 418.721–722.

D'Arcy Thompson (1961). On Growth and Form. Retrived from

http://dx.doi.org/10.1017/CBO9781107325852

David Kumar MS, Mohite PM, Sudhir Kamle, (2014). Structural Dynamic Analysis of

BioInspired Carbon Fiber/Polythene MAV Wings. Paper presented at the Recent

Advances in Mechanical Engineering, 2014. India, IIT.

de Smith, Goodchild, and Longley P.A. (2007) Geospatial Analysis: A comprehensive

Guide to Principles, Techniques and Software Tools. Retrieved from

http://www.spatialanalysisonline.com/HTML/index.htm

Dickinson MH, Lighton JRB. (1995). Muscle efficiency and elastic storage in the flight

motor of Drosophila. Science. 268. 87-89.

Draganfly Innovations Inc. (2015). An Introduction to Unmanned Aerial Vehicles.

Retrieved from http://www.draganfly.com/news/2008/08/24/introduction-to-

unmanned-aerial-vehicles-uavs/

Dumont E.R, I. R. Grosse, and Slater G.J. (2009). Requirements for comparing the

performance of finite element models of biological structures. Journal of

Theoretical Biology.256, 96-103.2009.

Ellington CP (1984a). The aerodynamics of insect flight. VI. Lift and power requirements.

Phil Trans Royal Society of London B.305.145-181.

Univers

ity of

Mala

ya

104

Ellington CP. (1984b). The aerodynamics of insect flight III. Kinematics. Proc Royal

Society of London B. 305.41-78.

Ellington CP. (1985). Power and efficiency of insect flight muscle. Journal of

Experimental Biology. 115.293-304.

Ellington CP. (1999). The novel aerodynamics of insect flight: application to micro-air

vehicles. Journal of Experimental Biology.202.3439-3448.

Federal Aviation Administration. (2015). Unmanned Aircraft Systems. Retrieved from

http:// www.faa.gov/uas/

Fenelon M.A.A., Furukawa T. Design of an Active Flapping Wing Mechanism and a Micro

Aerial Vehicle Using a Rotary Actuator. (2009). Mechanism and Machine Theory

45 (2009): 137–46.

Forman. R.T.T. (1987). The ethics of isolation, the spread of disturbance. and landscape

ecology. In: M.G. Turner (Editor). Landscape Heterogeneity and Disturbance.

Springer-Verlag, New York. 2 13-229.

Francis EH Tay, van Kan JA, Watt F, Choong WO. (2001). A novel micro-machining

method for the fabrication of thick-film SU-8 embedded micro-channels. Journal of

Micromechanics and Microengineering.11.27-32.

Frolov YS. (1975). Measuring the shape of geographical penomena: A history of the issue.

Soviet Geography Review and Translation. 16. 676-687.

Gardner, R.H., O’Neill, R.V., Turner, M.G. and Dale. V.H. (1989). Quantifying scale-

dependent effects of animal movement with simple percolation models. Landscape

Ecology. 3(2) 1l-227.

Gardner. R.H.. O’Neill. R.V. and Turner, M.G. (1993). Ecological implications of

landscape fragmentation. In: S.T.4. Pickett and M.J. McDonnell (Editors), Humans

as Components of Ecosystems: Subtle Human Effects and the Ecology of Populated

Univers

ity of

Mala

ya

105

Areas. Springer-Vcrlag. New York. 208-226.

Gorman J. (2003, June) What really happens when fruit flies fly? The New York Times.

Grodnitsky DL. (1999). Form and function of insec twings: the evolution of biological

structures. John Hopkins University Press. Baltimore.

Gui, L., Fink, T., Cao, Z., Sun, D., Seiner, J. M., & Streett, D. A. (2010). Fire Ant Alate

Wing Motion Data and Numerical Reconstruction. Journal of Insect Science, 10,

19. http://doi.org/10.1673/031.010.1901.

Gustafson, E.J. and Parker, G.R., (1992). Relationship between landcover proportion and

indices of landscape spatial pattern. Landscape Ecology. 7: 101-I 10.

Gustafson, E.J., Parker, G.R. and Backs, SE. (1994). Evaluating spatial pattern of wildlife

habitat: a case study of the WildTurkey (Meleagrisgallopa~o). Am. Midl. Nat. 131.

4-33.

Harris. L. (1984). The Fragmented Forest: Island-biogeography Theory and the

Preservation of Biotic Diversity. University of Chicago Press, Chicago, 211.

Herbert PGY R.C, Smith C.W, Wooton R.J, Evans K.E. (2000) The hindwing of the desert

locust (Schistocerca Gregaria Forskal): Finite Element Analysis of a Deployable

Structure. Journal of Experimental Biology. 203, 2945-2955.

Hisayoshi Naka YS, Hiromu Hashimoto.(2014, August 31 - September 3). Development of

the Artificial Wing Suitable for Flapping Micro Air Vehicle Based on Dragonfly

Wing. Paper presented at International Conference on Design Engineering and

Science, ICDES 2014, Pilsen, Czech Republic, Tokai University.

HuaiHui Ren, Xishu W, XuDong L, Ying Long C. (2013) Effects of dragonfly wing

structure on the dynamic performances. Journal of Bionic Engineering, 10, 28-38.

HuaiHui Ren, Xishu W, Ying Long C, Xu Dong L. (2012). Biomechanical behaviours of

dragonfly wings: Relationship between configuration and deformation. Chinese

Univers

ity of

Mala

ya

106

Physics B. (21).

Hwan Ko, Kim, J. Hong, Y.Yoo, Y. Lee, T.Lie Jin, H. Cheol Park, N. Seo Goo, D. Byun.

(2012). Micro/nanofabrication for a realistic beetle wing with a superhydrophobic

surface, Bioinspiration and Biomimetics, 7(1), 1748-3190

Jane Wang Z. (2004). The role of drag in insect hovering. Journal of Experimental

Biology.207(23). 4147-4155.

Jerzy Lewitowicz G.K, Krzysztof Sibilski, Josef Zurek. (2008). Modeling and simulation of

flapping wings micro-aerial vehicles flight dynamics. Paper presented at the

International Congress of the Aeronautical Sciences. Alaska, USA.

Jianyang Z, Chaoying Z, Chao W, Lin J. (2014). Effect of flexibility on flapping wing

characteristics under forward flight. Fluid Dynamics Research. 46(5), 055515.

Jiyu Sun and Bharath Bushan. (2012).The structure and mechanical properties of dragonfly

wings and their role on flyability. Comptes Rendus Mecanique, 340, 3-17.

Jongerius S.R and Lentink D. (2010). Structural Analysis of a Dragonfly Wing

Experimental Mechanics (50). 1323-1334.

Joong-Yeon Cho G.K, Sungwook Kim, Heon Lee.(2013).Replication of surface nano-

structure of the wing of dragonfly (Pantala Flavescens) using nano-molding and UV

nanoimprint lithography. Electronic Materials Letter. 9, 523-526.

Julian F.V Vincent, Olga A Bogatyreva, Nikolaj R Bogatyrev, Adrian Bowyer, Anja

Karina Pahl, S. (2006). Biomimetics: its practice and theory. The Royal Society,

3(9), 1742-5662.

Knaapen. J.P., Scheffer. M. and Harms. B., (1992). Estimating habitat isolation in

landscape planning. Landscape Urban Plann., 23: l-16.

Krohne. D.T.. Dubbs. B.A. and Baccus, R. (1984). An analysis of dispersal in an

unmanipulated population of Peromyscus leucopus. Am. Midl. Nat., 112. 146-156.

Univers

ity of

Mala

ya

107

Kutsch, W., Martz, H. and Gäde, G. (2002), Flight capability and flight motor pattern in a

sedentary South African grasshopper, Phymateus morbillosus– a comparison with

migratory species. Physiological Entomology, 27: 39–50. doi: 10.1046/j.1365-

3032.2002.00267.x

Kyriakides S, Arseculeratne R, Perry E.J, Liechti K.M. (1995). On the compressive failure

of fiber reinforced composites. International Journal of Solid Structures. 32. 685-

738.

Leban M.B and Tisu R. (2013) The effect of TiN inclusions and deformation-induced

martensite on the corrosion properties of AISI 321 stainless steel. Engineering

Failure Analysis. 33(10),430-438.

Lehman F-O, Sane SP, Dickinson MH (2005). The aerodynamic effects of wing-wing

interaction in flapping insect wings. Journal of Experimental Biology. 208.2075-

3092.

Lillesand. T.M. and Kiefer. R.W., (1987). Remote Sensing and Image Interpretation.

Wiley, New York, 721.

Lingxiao Zheng TL, Hedrick, Rajat Mittal.(2013). Time-varying wing twist improves

aerodynamic efficiency of forward flight in butterflies. PLOS ONE. 8(1).005306

Lord. J.M. and Norton, D.A., 1990. Scale and the spatial concept of fragmentation.

Conserv. Biol., 4. 197-202.

Luo H, Tian, Fangbao, Song, Jialei, Lu, Xi-Yun. (2012). Aerodynamic cause of the

asymmetric wing deformation of insect wings. APS Division of Fluid Dynamics

E32.003.

Luttges MW. 1989. Accomplished insect fliers. In: Gad-el –Hak M (ed). Frontiers in

experimental fluid mechanics, Lecture noted in Engineering, Springer, Berlin, 429-

456.

Univers

ity of

Mala

ya

108

Marcin Luczak, S.M., Bart Peeters, Kim Branner, Peter Berring, Maciej Kahsin. (2011).

Dynamic Investigation of Twist-Bend Coupling in a Wind Turbine Blade.

Theoretical And Applied Mechanics (49). 765-789.

Maria Mingallon S.R. (2011).The Architecture of the Dragonfly Wing: A Study of the

Structural and Fluid Dynamic Capabilities of the Anisoptera's Forewing. Paper

presented at the International Mechanical Engineering Congress & Exposition,

Denver, Colorado, USA.

Maybury WJ, Lehmann F.O. (2004). The fluid dynamics of flight control by kinematic

phase lag variation between two robotic insect wings. Journal of Experimental

Biology. 207.4707-4726.

McDonald M, Agrawal Sunil K. (2010). Design of a Bio-Inspired Spherical Four-Bar

Mechanism for Flapping-Wing Micro Air-Vehicle Applications. Journal of

Mechanisms and Robotics. 021012.

McMichael J.M. and Francis, M. (1997, July). Micro Air Vehicles-Towards a new

dimension in flight. Federation of American Scientists. Retrieved from

http://www.fas.org/irp/program/collect/docs/mav_auvsi.htm

Michelson, R.C. and Reece, S., (1998) March. Update on flapping wing micro air vehicle

research-ongoing work to develop a flapping wing, crawling entomopter. In 13th

Bristol International RPV/UAV Systems Conference Proceedings, Bristol

England (Vol. 30, pp. 30-1).

Miller LA, Peskin CS. (2004). When vortices stick: an aerodynamic transition in tiny insect

flight. Journal of Experimental Biology. 207. 3073-3088.

Mountcastle AM. (2009). Aerodynamic and functional consequences of wing compliance.

Journal of Experimental Fluids. 46, 873-82.

Newman et al. (1986). An approach to the mechanics of pleating in dragonfly wings.

Journal of Experimental Biology. 125, 361-72.

Ngoc San Ha and Nam Seo Goo. (2013 January). Flapping frequency and resonant

frequency of insect wings. Paper presented at the 10th International Conference on

Ubiquitous Robots and Ambient Intelligence (URAI), Konkuk University.

Univers

ity of

Mala

ya

109

doi:10.1109/URAI.2013.6677463

Nguyen QV, Park H, Byun D, Goo N. (2010, September 19-23). Recent progress in

developing a beetle-mimicking flapping-wing system. Paper presented at the World

Automation Congress 2010, Kobe. US, IEEE.

Noms S (2001). Taking Flight. Oxford Today.14(1)

Norberg RA (1975). Hovering flight of the dragonfly Aeschna juncea. Wu TY-T,Brokaw

CJ, Brennen C (eds) Kinematics and Aerodynamics.2 Plenum Press, New York.

763-781

Osei-Antwi M, J. de Castro, Vassilopoulos A.P, and Keller T. (2013). Shear mechanical

characterization of balsa wood as core material of composite sandwich panels.

Construction and Building Materials. 41(4). 231-238.

Osei-Antwi M, J. de Castro, Vassilopoulos A.P, and Keller T. (2014). Fracture in complex

balsa cores of fiber-reinforced polymer sandwich structures. Construction and

Building Materials. 71,194-201.

Rahman. (2012). Assessment of Bridge Service Life using Wireless Sensor Network.

Youngstown State University, Department of Graduate Studies and Research.

Rajabi M.M.H and Darvizeh A. (2011). Investigation of microstructure, natural frequencies

and vibration modes of dragonfly wing. Journal of Bionic Engineering. 8. 165-173.

Reavis MA, Luttges MW. (1988). Aerodynamic forces produced by a dragonfly. AIAA J.

88-0330. 1-13

Rieger N.F. (1999). The Relationship Between Finite Element Analysis and Modal

Analysis. Stress Technology Incorporated. Retrieved from http://www.sti-

tech.com/dl/feapaper.pdf

Rose E. Hutchins (n.d). Insects. Retrieved from Scholastic website

http://teacher.scholastic.com/activities/explorations/bug/libraryarticle.asp?ItemID=1

Univers

ity of

Mala

ya

110

12&SubjectID=110&categoryID=4

Rubentheren V, T. A. Ward, C. Y. Chee, and C. K. Tang. (2015). Processing and analysis

of chitosan nanocomposites reinforced with chitin whiskers and tannic acid as a

crosslinker. Carbohydrate Polymers.115.379-387.

Rupan Talucdher and Kunigal Shivakumar. (2013). Tensile Properties of Veins of

Damselfly Wing. Journal of Biomaterials and Nanobiotechnology. 4. 247-255.

Rüppell G (1989). Kinematic analysis of symmetrical flight manoeuvres of Odonata.

Journal of Experimental Biology. 144.13-42

Saharon D, Luttges MW. (1987). Three dimensional flow produced by a ptiching-plunging

model dragonfly wing. AIAA J. 87-0121. 1-17

Shang J.K, Combes S, Finio B.M, Wood R.J. (2009). Artificial insect wings of diverse

morphology for flapping-wing micro air vehicles. Bioinspiration and Biomimetics,

4(3), 1748-3190.

Shyy W, Aono H, Chimakurthi S.K, Trizila P, Kang C.K, Cesnik, et al. (2010) Recent

progress in flapping wing aerodynamics and aeroelasticity. Progress in Aerospace

Sciences,46, 284-327.

Shyy W., Berg M., Ljungqvist D. (1999). Flapping and flexible wings for biological and

micro air vehicles. Progress in Aerospace Sciences. 35. 455–505.

Smithsonian Institution (2010). Bug Info: Insect Flight. Encyclopedia Smithsonian Online

Retrieved from http://www.si.edu/Encyclopedia_SI/nmnh/buginfo/insflght.htm

Sudo S, Tsuyuki K, Ito Y, Tani J. (2001) The wing apparatus and flapping behavior of

hymenoptera. JSME International Journal, Series C. 44. 1103–1110.

Sudo S, Tsuyuki K, Kanno K. (2005). Wing characteristics and flapping behavior of flying

Insects.Experimental Mechanics. 45.550–555.

Univers

ity of

Mala

ya

111

Sun M, Tang J (2002). Unsteady aerodynamics force generation by a model fruit fly wing

flapping motion. Journal of Experimental Biology. 205. 55-70.

Tamilarasan, Uthirapathy, Karunamoorthy, Loganathan, & Palanikumar, Kayaroganam.

(2015). Mechanical Properties Evaluation of the Carbon Fibre Reinforced

Aluminium Sandwich Composites. Materials Research, 18(5), 1029-

1037.https://dx.doi.org/10.1590/1516-1439.017215

Tanaka H. and Wood R.J. (2010). Fabrication of corrugated artificial insect wings using

laser micromachined molds. Journal of Micromechanics and Microengineering.

20075008.

The UAV. (2015). The UAV. Retrieved from http://www.theuav.com/#31

Tobalske B.W. (2000). Physiology and Biomechanics of Gait Selection in Flying Birds.

Physiol Biochem Zoology. 73(6):736–750.

Tobalske B.W., Dial K.P. (1996). Flight kinematics of Black-billed Magpies and pigeons

over a wide range of speeds. Journal of Experimental Biology. 99: 263–280.

Tu Delft University of Technology. (2013). Robot Dragonfly DelFly Explorer avoids

obstacles by itself. Retrieved from http://www.

tudelft.nl/en/current/latestnews/article/detail/robotlibelle-delfly-explorer-ontwijkt-

zelf-obstakels/

Usherwood JR, Ellington CP. (2002). The aerodynamics of revolving wings I. Model

hawkmoth wings. Journal of Experimental Biology. 205. 1547-1564.

Vanderberghe N, Zhang J, Childress S. (2004). Symmetry breaking leads to forward

flapping flight. Journal of Fluid Mechanics. 506.147-155.

Vogel S. (2013) Comparative Biomechanics: Life's Physical World (2nd Edition).

Princeton University Press.

Wakeling J.M and Ellington C.P. (1997). Dragonfly Flight: Velocities, Accelerations and

Univers

ity of

Mala

ya

112

Kinematics of Flapping Flights. Journal of Experimental Biology.200, (3) 557-582.

Ward, T. A., Rezadad, M., Fearday, C. J., & Viyapuri, R. (2015). A Review of Biomimetic

Air Vehicle Research: 1984-2014. International Journal of Micro Air Vehicles,

7(3), 375-394.

Warren A.D, Harniman R.L, Collins A.M, Davis S.A, Younes C.M, Flewitt P.E.J, et al.

(2015). Comparison between magnetic force microscopy and electron back-scatter

diffraction for ferrite quantification in Type 321 stainless steel. Ultramicroscopy,

148 (1), 1-9.

Weis-Fogh T (1972). Energetics of hovering flight in hummingbirds and Drosophila.

Journal of Experimental Biology. 56.79-104.

Wootton RJ.(1981). Support and deformability in insect wings. Journal of Zoology

193:447-68.

Wootton RJ.(1993) Leading edge section and asymmetric twisting in the wings of flying

butterflies (Insecta Papilionoidea). Journal of Experimental Biology. 180. 105-17.

Xinyan Deng and Zheng Hu. (2008). Wing-wing interactions in dragonfly flight. The

Neuromorphic Engineer. 10.2417/1200811.1269.

Xue Li Z, W. Shen, G. Shu Tong, Jia-meng T, Loc VQ. (2009). On the vein stiffening

membrane structure of a dragonfly hindwing. Journal of Zheijang University. 10,

72-81.

Yang W Q, Song W P. (2012). The effects of span-wise and chord-wise flexibility on the

aerodynamic performance of micro flapping-wing. Chinese Science Bulletin. 57.

2887-97.

Yirka, Bob. (2013, April) Festo builds BionicOpter-fully functional robot dragonfly.

Phys.Org. Retrieved from http://phys.org/news/2013-04-festo-bionicopterfully-

functional-robot-dragonfly.html

Young J WS, Bomphrey RJ, Taylor GK, Thomas ALR. (2009). Details of insect wing

design and deformation enhance aerodynamic function and flight efficiency.

Univers

ity of

Mala

ya

113

Science 325, 1549–52.

Zhao L, Huang Q, Deng X, Sane SP.(2009). Aerodynamic effects of flexibility in flapping

wings. Journal of The Royal Society Interface. 12(113)1742-5662.

Zheng Hu and Xinyan Deng (2009). A Flight Strategy for Intelligent Aerial Vehicles

Learned from Dragonfly, Aerial Vehicles, Retrieved from:

http://www.intechopen.com/books/aerial_vehicles/a_flight_strategy_for_intelligent

_aerial_vehicles_learned_from_dragonfly.

Univers

ity of

Mala

ya

114

LIST OF PUBLICATIONS

Journal articles

[1] Praveena N. Sivasankaran, Thomas A. Ward (2015). Spatial network analysis to

construct simplified wing structural models for Biomimetic Micro Air Vehicles.

Aerospace Science and Technology.(Q1 Journal:1.000 IF)

[2] Nair, P., Ward, T., Rubentheren, V., Mohd Rafie J. (2015). Static Strength Analysis

of Dragonfly Inspired wing for Biomimetic Micro Air Vehicles. Chinese Journal of

Aeronautics. (Accepted for publication) (Q1 Journal:1.070 IF)

[3] Praveena N. Sivasankaran, Thomas A. Ward, Erfan Salami, Rubentheren Viyapuri,

Christopher Fearday. (2015). An experimental study of the aeroelastic properties of

dragonfly-like flapping wings for use in Biomimetic Micro Air Vehicles (BMAV).

Chinese Journal of Aeronautics (Accepted for publication) (Q1 Journal: 1.070 IF)

[4] Rubentheren, V., Ward, T. A., Chee, C. Y., & Tang, C. K. (2015). Processing and

analysis of chitosan nanocomposites reinforced with chitin whiskers and tannic acid

as a crosslinker. Carbohydrate Polymers, 115, 379-387. (Q1 Journal: 4.074 IF)

[5] Rubentheren, V., Ward, T., Chee, C., & Nair, P. (2015). Physical and chemical

reinforcement of chitosan film using nanocrystalline cellulose and tannic acid.

Cellulose, 22(4), 2529-2541. (Q1 Journal:3.573 IF)

[6] Rubentheren, V., Ward, T., Chee, C., Nair, P., Salami, E., & Fearday, C. (2015).

Effects of heat treatment on chitosan nanocomposite film reinforced with

nanocrystalline cellulose and tannic acid. Carbohydrate Polymers. (Accepted with

minor revision) (Q1 Journal: 4.074 IF)

Univers

ity of

Mala

ya

115

Conferences

[1] Nair, P., Ward, T., Rubentheren, V., Johan, M., R. (2015). Experimental analysis of

artificial dragonfly wings using black graphite and fibreglass for use in Biomimetic

Micro Air Vehicles (BMAV). MATEC Web of Conferences 30,03001.

[2] Rubentheren, V., Ward, T., Chee, C., Nair, P., Erfan, S., (2015). Development of

heat treated chitosan nano-composite film for the wing membrane of biomimetic

micro air vehicle (BMAV). European conference on organized film ECOF 14.

Univers

ity of

Mala

ya

116

APPENDICES

Appendix A: High speed camera images of PLA wing frame models

(a) Side view of PLA wing frame

(b) Front view of PLA wing frame

Univers

ity of

Mala

ya

117

Appendix B: High speed camera images of acrylic wing frame models

(a) Side view of acrylic wing frame

(b) Front view of acrylic wing frame

Univers

ity of

Mala

ya

118

Appendix C: High speed camera images of ABS wing frame models

(a) Side view of ABS wing frame

(b) Front view of ABS wing frame

Univers

ity of

Mala

ya